# EasyJev Docs — full text # Generated 2026-11-17. Source of truth: https://docs.easyjev.online (per-page .md twins follow the same order) # Live data (prices, model list) via https://api.easyjev.online/v1/models or MCP https://api.easyjev.online/mcp # Documentation Home > EasyJev documentation: integrate the System One decision models (Jev, pplx-decider, Drex, d1, Clef and more) in five minutes — official typesafe-sdk compatible, one base_url swap. --- # Quickstart > Three steps to your first judgment call: sign in with an email code, create an sk- key, swap base_url and api_key in the official typesafe-sdk — done in five minutes. ## Three steps to your first call > **1 · Sign in with an email code** > Open the [EasyJev console](https://www.easyjev.online/#/login), enter your email, and type the 6-digit code from your inbox. The first sign-in creates the account automatically — no separate registration, no password, and new accounts receive a small free credit to try the API. > **2 · Create an sk- API key** > In the console open **API Keys** and click create. The plaintext key (`sk-...`) is shown **only once** — copy it immediately. Only a SHA-256 hash is stored server-side, so a lost key cannot be recovered (just create a new one). See [API Keys](https://docs.easyjev.online/keys/). > **3 · Swap base_url and call** > Point the official `typesafe-sdk` at EasyJev: change `base_url` to `https://api.easyjev.online` and `api_key` to your sk- key. Nothing else in your code changes — request shapes, error classes and `request_id` tracing stay on the official contract. ## Base URL All API requests go to this single origin. It never changes per model — pick models inside the request body. ```base_url https://api.easyjev.online ``` ## Official SDK (Python) ```python from typesafe_sdk import Noul, TypeSafeClient # The only two edits: your sk- key from this site, and our base_url client = TypeSafeClient( api_key="sk-your-key", base_url="https://api.easyjev.online", ) result = client.system_one( state="User asks: can I get a refund for order #1024?", questions={"refund": Noul(instructions="Is this a refund request?")}, ) print(result.nouls["refund"].noul) # probability of "yes", e.g. 0.97 ``` ### Raw HTTP (curl) ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "typesafe/jev-latest", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` ## Where to go next - [Python SDK guide](https://docs.easyjev.online/sdks/python/) — all three question types, error mapping, multimodal assets - [POST /v1/systemone reference](https://docs.easyjev.online/api/systemone/) — every field and semantics - [Model catalog](https://docs.easyjev.online/models/) — every decision model on sale, with live prices - [GET /v1/balance](https://docs.easyjev.online/api/balance/) — check your remaining credit from code - [Agents & MCP](https://docs.easyjev.online/agents/) — let your AI agent read these docs itself --- # API Keys > Create, list, disable and delete sk- API keys in the EasyJev console. The plaintext key is shown only once at creation; only SHA-256 hashes are stored. ## Managing keys in the console Open the [API Keys](https://www.easyjev.online/#/keys) page in the console. You can create, rename, disable and delete keys; each key is listed with its prefix, creation time and last-use time. There is a per-account key limit (shown on the page) — delete unused keys to free slots. ## Security model | Rule | Detail | | --- | --- | | Plaintext shown once | The full `sk-...` appears exactly once at creation. Copy it then — there is no “show again”. | | Hashed at rest | Only SHA-256 hashes are stored. A lost key cannot be recovered — create a new one and delete the old. | | Per-key scoping | Disable a key without touching others; rotate by creating the new key first, switching callers, then deleting the old. | | Rate limits per key | RPM/TPM limits are enforced per key, so one hot service cannot starve your other integrations. | ## Default rate limits Every key gets a sliding-window limit — by default **600 requests/minute** and **4,500,000 tokens/minute** (the console shows your live values). Hitting either returns `429`; media bytes are excluded from the estimate and the window frees capacity continuously, so a short backoff is enough. Need more? Contact support with your use case. --- # Billing & Pricing > How EasyJev bills: integer-quota ledger (1,000,000,000 quota = $1), per-model input pricing in USD per million tokens, free output tokens, failed calls never charged, top-up options. ## The quota unit All balances and prices are tracked in integer **quota**: `1,000,000,000 quota = $1 USD`. Integer arithmetic end-to-end means no floating-point drift — ever. The single settlement currency is USD. ## Per-model prices The table below is rendered live from [GET /v1/models](https://docs.easyjev.online/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. | Model | Input $/M tokens | Output $/M tokens | | --- | --- | --- | ## How a call is billed | Rule | Detail | | --- | --- | | Metered input | Price is per million input tokens, per model; settle exactly against the upstream-reported `usage.input_tokens`. | | Output free | System One models charge nothing for output tokens across the current catalog. | | Failures are free | 4xx/5xx/timeouts never bill — only 200 responses settle. | | Negative-then-block | A delivered call may take the balance slightly negative (delivered service must be accounted); the next call is then blocked with 402 until you top up. | | Live balance from code | Poll [GET /v1/balance](https://docs.easyjev.online/api/balance/) (free) and configure a low-balance alert in the console. | ### Topping up Top up in the [console recharge page](https://www.easyjev.online/#/recharge): online card / Apple Pay / Google Pay / WeChat Pay via our payment provider, or contact support for manual transfer during rollout. New accounts get a small free credit on first sign-in (amount is a runtime setting — the console shows the live value). --- # Python · typesafe-sdk > Use the official typesafe-sdk Python package against EasyJev: pip install, TypeSafeClient with your sk- key and our base_url, Noul/choice/score questions, error mapping and request_id tracing. ## Install ```shell pip install typesafe-sdk ``` ## Create the client ```python from typesafe_sdk import TypeSafeClient client = TypeSafeClient( api_key="sk-your-key", # EasyJev console -> API Keys base_url="https://api.easyjev.online", # the only other edit ) ``` Both values can also come from environment variables: `TYPESAFE_API_KEY` and `TYPESAFE_BASE_URL`. ## The three question types **Noul (yes/no)** — returns a calibrated probability of “yes” in [0, 1]. Best for gates: is this a refund request? is this spam? ```python from typesafe_sdk import Noul result = client.system_one( state="User asks: can I get a refund for order #1024?", questions={"refund": Noul(instructions="Is this a refund request?")}, ) print(result.nouls["refund"].noul) # probability of "yes", e.g. 0.97 ``` **Choice (single pick)** — pick one named candidate from up to 255 options with per-candidate probabilities. Ideal for routing and classification. ```python from typesafe_sdk import Choice result = client.system_one( state="Ticket: my order #1024 arrived damaged", questions={ "route": Choice( instructions="Pick the team that should handle this ticket", criteria={ "billing": "Refunds, charges, payment issues", "support": "Order status, delivery, damaged items", "sales": "Pre-sale questions, quotes", }, ) }, ) print(result.choices["route"].choice) # e.g. "support" ``` **Score (ordinal)** — score against 2–10 ordered levels and get a value that interpolates between them. Ideal for severity and quality rubrics. ```python from typesafe_sdk import Score result = client.system_one( state="Comment: this product changed my life, best purchase ever!!", questions={ "severity": Score( instructions="Rate how urgent moderation is", criteria=["benign", "minor", "review", "block"], ) }, ) print(result.scores["severity"].score) # e.g. 0.25 (between benign and minor) ``` ## Error handling The SDK raises its official exception classes unchanged; EasyJev maps every gateway error onto the same class hierarchy you would see against the official API. Every response also carries `request_id` (header `X-Request-Id`) — include it when contacting support. ```python from typesafe_sdk import errors try: result = client.system_one( state="...", questions={"q": {"type": "noul", "instructions": "..."}}, model="typesafe/jev-latest", ) except errors.AuthenticationError: ... # 401/403: wrong or disabled sk- key except errors.RateLimitError: ... # 429: RPM/TPM limit hit, back off and retry except errors.APIError as e: ... # other gateway errors; e carries the unified error body ``` --- # HTTP & curl > Call POST /v1/systemone directly over HTTP from any language: curl, Python requests, Node fetch and Go examples with the exact JSON request/response shapes. ## The endpoint Every judgment is one `POST` to `https://api.easyjev.online/v1/systemone` with a Bearer sk- key. Auth header: `Authorization: Bearer sk-...`. The JSON body needs `model`, `state` and `questions` — unknown fields pass through to the upstream unchanged. ## Examples in four languages ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "typesafe/jev-latest", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` ```python import requests resp = requests.post( "https://api.easyjev.online/v1/systemone", headers={"Authorization": "Bearer sk-your-key"}, json={ "model": "typesafe/jev-latest", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} }, }, timeout=30, ) resp.raise_for_status() print(resp.json()["answers"]["refund"]["noul"]) # e.g. 0.97 ``` ```javascript const resp = await fetch("https://api.easyjev.online/v1/systemone", { method: "POST", headers: { Authorization: "Bearer sk-your-key", "Content-Type": "application/json", }, body: JSON.stringify({ model: "typesafe/jev-latest", state: "User asks: can I get a refund for order #1024?", questions: { refund: { type: "noul", instructions: "Is this a refund request?" }, }, }), }); const data = await resp.json(); console.log(data.answers.refund.noul); // e.g. 0.97 ``` ```go package main import ( "bytes" "encoding/json" "fmt" "net/http" ) func main() { payload, _ := json.Marshal(map[string]any{ "model": "typesafe/jev-latest", "state": "User asks: can I get a refund for order #1024?", "questions": map[string]any{ "refund": map[string]any{ "type": "noul", "instructions": "Is this a refund request?", }, }, }) req, _ := http.NewRequest("POST", "https://api.easyjev.online/v1/systemone", bytes.NewReader(payload)) req.Header.Set("Authorization", "Bearer sk-your-key") req.Header.Set("Content-Type", "application/json") resp, err := http.DefaultClient.Do(req) if err != nil { panic(err) } defer resp.Body.Close() var out struct { Answers map[string]struct { Noul float64 `json:"noul"` } `json:"answers"` } json.NewDecoder(resp.Body).Decode(&out) fmt.Println(out.Answers["refund"].Noul) // e.g. 0.97 } ``` ## Reading the response Answers arrive keyed by question id under `answers`, each with the question `type` echoed plus type-specific fields (`noul` probability, `choice` + `probabilities`, interpolated `score`, `confidence`). Billing meters appear under `usage`. See the [full reference](https://docs.easyjev.online/sdks/systemone/) for every field. ```json { "model": "typesafe/jev-latest", "answers": { "refund": { "type": "noul", "noul": 0.97, "confidence": 0.93 } }, "usage": {"input_tokens": 1234, "output_tokens": 0} } ``` --- # Images & Multimodal Assets > Send images to decision models: the official SDK Image asset (multipart) or raw JSON inline base64, the 50MB per-file limit, automatic TypeSafe-Danger-Zone-Images opt-in, and per-model vision constraints. ## Two ways to attach an image ### Official SDK (multipart, recommended) Pass the SDK’s `Image` asset in `assets` and reference it inside `state`/`instructions`/`criteria`. The SDK encodes the request as `multipart/form-data`; EasyJev translates it to the inline form the upstream accepts and delivers the bytes intact — media never touch the database. ```python from typesafe_sdk import Image, Noul photo = Image("product.png") result = client.system_one( state=f"Product photo: {photo}", questions={"damaged": Noul(instructions="Is the product damaged?")}, model="typesafe/jev-latest", # use a vision-capable model assets={"photo": photo}, ) ``` ### Raw JSON (inline base64) Without the SDK, put the asset directly in the JSON body under `assets` as `{"key": {"type": "image", "format": "png", "data_base64": "..."}}` and reference the key the same way. Both forms share the same limits. ```json { "model": "typesafe/jev-latest", "state": {"photo": ""}, "questions": { "damaged": {"type": "noul", "instructions": "Is the product damaged?"} }, "assets": { "photo": {"type": "image", "format": "png", "data_base64": "iVBORw0KGgo..."} } } ``` ## Limits & semantics | Item | Rule | | --- | --- | | Per media file | 50 MB max (multipart: per-part bytes; JSON: decoded base64 size). Over the limit → 413 `media_too_large`. | | Total request body | multipart 60 MB / pure JSON 70 MB defaults (must hold a 50 MB media after base64 expansion). | | Billing | Images bill as ordinary input tokens by upstream usage — no media surcharge; rate-limit estimation excludes media bytes. | | Opt-in header | The official `TypeSafe-Danger-Zone-Images: true` header is added automatically when assets are present — you never need to set it by hand. | | Privacy | Media bytes are never stored: inline bodies are redacted to `` placeholders before archiving, multipart archives only the JSON part (truncated ~64 KB). | > **Vision support varies by model.** In practice the Jev family accepts images (smallest edge ≈ 512 px; references can be embedded in long text), while some non-Jev providers only accept whole-string references as top-level state items. Unsupported combinations return an upstream 4xx unchanged and are never billed. Pick a vision-capable model such as `typesafe/jev-latest` when sending images. --- # POST /v1/systemone > Complete reference for the judgment endpoint: model/state/questions fields, the three question types (noul, choice, score), processing semantics, response shape and pass-through upstream errors. ## Request fields | Field | Type | Required | Description | | --- | --- | --- | --- | | model | string | Yes | Model name as returned by [GET /v1/models](https://docs.easyjev.online/api/models/), e.g. `typesafe/jev-latest`. Legacy bare names (`jev-latest`, `jev-1.13.0`) still work via alias normalization. | | state | string / object / array | Yes | The content being judged. Free-form: a question, a ticket, a product record — anything the questions refer to. | | questions | object | Yes | Map of `{question_id: question_def}`. At least one entry; ids are echoed back in the response. | | (others) | any | No | Unknown fields pass through to the upstream unchanged (forward compatibility). | ```json { "model": "typesafe/jev-latest", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"}, "urgency": { "type": "score", "instructions": "How urgent is this?", "criteria": ["low", "normal", "high"] } } } ``` ## Question types | Type | type | Required fields | Returns | | --- | --- | --- | --- | | Yes/no | `noul` | — | Probability of “yes” in [0,1]; optional `criteria.true/false` descriptions. | | Single choice | `choice` | `criteria` | Map of option name → description (≤255 options, description may be null); returns the pick plus per-option `probabilities`. | | Ordinal score | `score` | `criteria` | Ordered array of 2–10 levels; returns a value that interpolates between levels. | `instructions` is optional on every question type and accepts a string, object or array. Images can be attached via `assets` — see [Images & Multimodal Assets](https://docs.easyjev.online/sdks/multimodal/). ## Processing semantics (in order) 1. **1.** Request-size limits (single media 50 MB; total: multipart 60 MB / JSON 70 MB) → 413 2. **2.** Minimal structure check — missing model/state/questions → 422 3. **3.** Model whitelist (after alias normalization) — unknown model → 404, not forwarded 4. **4.** Per-key rate limits (default 600 RPM / 4,500,000 TPM) → 429 5. **5.** Balance pre-check — zero or negative balance → 402 (a call may settle into the negative once) 6. **6.** Upstream forwarding with failover: up to 3 channel+key attempts (401 auto-disables a key, 429 cools down, 5xx/network errors fail over) 7. **7.** Billing on 200 only: input tokens × input price + output tokens × output price, settled exactly against upstream usage; failed calls are free 8. **8.** Every call (success or failure) is logged with a request_id returned in response headers ## Success response The 200 body is the upstream response passed through unchanged (unknown fields included). `usage` is the billing meter: only `input_tokens` / `output_tokens` / `cost` (when the upstream returns it) are exposed. ```json { "model": "typesafe/jev-latest", "answers": { "refund": { "type": "noul", "noul": 0.97, "confidence": 0.93 }, "urgency": { "type": "score", "score": 1.4, "confidence": 0.88 } }, "usage": {"input_tokens": 1234, "output_tokens": 0} } ``` ### Response headers Every response (success or error) carries `X-Request-Id` and `x-typesafe-request-id` with the same value — use it for reconciliation and when contacting support; the official SDK also surfaces it as `resp.request_id`. ### Upstream pass-through Upstream validation failures (`400` / `404` / `413` / `422`) are passed through **unchanged** — the body is the upstream’s error JSON, not the gateway envelope — and are never billed. All other failures use the [unified gateway error shape](https://docs.easyjev.online/api/errors/). --- # GET /v1/models > List every model on sale with live prices: name, description, release_date and per-million-token input/output prices in integer quota. No authentication required. ```curl curl https://api.easyjev.online/v1/models ``` ```json { "models": [ { "name": "typesafe/jev-latest", "description": "TypeSafe: Jev (latest)", "release_date": "2026-10-14", "input_price_quota_per_m": 42000000, "output_price_quota_per_m": 0 }, { "name": "perplexity/pplx-decider-v1.2", "description": "Perplexity: Decider V1.2", "release_date": "2026-10-16", "input_price_quota_per_m": 20000000, "output_price_quota_per_m": 0 } ] } ``` ## Response fields | Field | Description | | --- | --- | | name | Model name — exactly what goes into the `model` field of systemone | | description | Human-readable description (catalog remark) | | release_date | Release date, `YYYY-MM-DD` | | input_price_quota_per_m | Input price per million tokens, integer quota (1e9 quota = $1) | | output_price_quota_per_m | Output price per million tokens — 0 means free | > No authentication required (an Authorization header is ignored, not validated). The list mirrors the [model catalog page](https://docs.easyjev.online/models/); this endpoint is the live source of truth — catalog pages are build-time snapshots. MCP users can call `list_models` for the same data. --- # GET /v1/balance > Check the live account balance from your code with the same sk- key: balance in quota and USD, low-balance flag and your alert threshold. Free, no tokens consumed. Check your live balance from code with the same sk- key you call systemone with — no upstream call, no tokens consumed, no charge. The response also reports whether the balance is below your alert threshold (configured in the console). ```curl curl https://api.easyjev.online/v1/balance \ -H "Authorization: Bearer sk-your-key" ``` ```json { "balance_quota": 420000000, "balance_usd": "0.42", "low_balance": false, "alert_threshold_usd": "1.00" } ``` ## Field · /v1/balance | Field | Description | | --- | --- | | balance_quota | Balance in integer quota (1e9 quota = $1) | | balance_usd | Balance formatted as USD text | | low_balance | true when balance < your alert threshold (live fact, independent of the email switch) | | alert_threshold_usd | Your alert threshold formatted as USD text | --- # Errors & Status Codes > Every status code EasyJev returns: the unified error JSON shape with request_id, the full code table (400/401/402/403/404/413/422/429/502), retry guidance and upstream pass-through semantics. ## The unified error shape Gateway-originated failures (auth, quota, rate limit, upstream exhaustion…) all return this JSON envelope on /v1/* paths. Upstream validation errors are the exception — they pass through unchanged (see the systemone page). ```json { "error": { "message": "Balance is not enough to make this call", "type": "insufficient_quota", "code": "insufficient_quota" }, "request_id": "8f14e45fceea167a5a36dedd4bea2543" } ``` ## Status codes | HTTP | type | code | Meaning / what to do | | --- | --- | --- | --- | | 400 | invalid_request_error | upstream pass-through | Upstream rejected the request body. Read the upstream message; not billed. | | 401 | authentication_error | invalid_api_key | Missing/wrong Bearer key. Check the sk- value; a disabled key also lands here. | | 402 | insufficient_quota | insufficient_quota | Balance is zero/negative. [Top up](https://docs.easyjev.online/billing/), then retry. | | 403 | authentication_error | account_suspended / key_disabled | Account or key disabled — contact support. | | 404 | invalid_request_error | model_not_found | Model not on sale (or typo). List valid names via GET /v1/models. | | 413 | invalid_request_error | payload_too_large / media_too_large | Request over size limits — shrink the body or split the asset. | | 422 | invalid_request_error | bad_request (gateway) / upstream pass-through | Missing model/state/questions, or upstream schema rejection. | | 429 | rate_limit_error | rate_limit_exceeded | Per-key RPM/TPM hit. Back off (the limit resets inside a minute) and retry. | | 502 | api_error | upstream_failed | All channel+key attempts failed after failover. Safe to retry with backoff; contact support with request_id if it persists. | ## Retry guidance > **Retry safely on** 429 (after backing off — the sliding window frees capacity within a minute) and 502 (failover was exhausted; transient upstream issues are common). **Do not blind-retry** 400/401/402/403/404/413/422 — the request itself must change first. Every retry carries a fresh request_id, so reconciliation stays clean. --- # Model Catalog > Every System One decision model on sale at EasyJev, with live pricing: Jev (typesafe/jev-*), pplx-decider, Drex, Liquid d1, Cloudflare Clef, Solar Decider, Span and more — one endpoint for the whole catalog. ## Featured models --- # Model Comparison > Compare System One decision models side by side: input price per million tokens, free output, vendors and release dates, with price-tier picks (cheapest, balanced, flagship) and one-line model switching code. ## Price matrix (live) Currently — models on sale — rendered live from the gateway. Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. | Model | Vendor | Input $/M | Tier | | --- | --- | --- | --- | ## Picks by price tier --- # Agents & MCP > EasyJev is agent-native: read docs via llms.txt or per-page .md twins, query live prices and search documentation through the MCP server at api.easyjev.online/mcp — wire any AI agent or coding assistant in minutes. ## Why agent-native A large share of EasyJev users never read docs in a browser — they send an AI agent. So this documentation site treats agents as first-class readers: every page is real HTML at a stable URL, has a Markdown twin (append `.md`), exposes a curated [llms.txt](https://docs.easyjev.online/llms.txt) index plus a full-text [llms-full.txt](https://docs.easyjev.online/llms-full.txt), and live data (prices, model list, free credit policy, doc search) is queryable through a read-only MCP server. No JS execution required anywhere. ## Read any page as Markdown Append `.md` to any English or Simplified-Chinese page URL and you get clean Markdown instead of HTML — ideal for fetch-and-read agents: ```bash curl https://docs.easyjev.online/api/systemone.md # Markdown twin of the systemone reference ``` ## The MCP server A read-only, unauthenticated Model Context Protocol server (Streamable HTTP) is mounted at `https://api.easyjev.online/mcp`. It exposes live data — never a stale snapshot — and absolutely no user data, channel secrets or write operations. Add it to any MCP client: ### Tools | Tool | Description | | --- | --- | | list_models | Every decision model on sale with live prices, vendors, release dates and page URLs. | | get_pricing | Live pricing and details for one model by exact name (e.g. "typesafe/jev-1.14"). | | get_free_credit_policy | Sign-up bonus (live value), billing unit, pay-as-you-go terms, top-up channels. | | get_model_comparison | The whole catalog as a price-sorted comparison with tier picks (cheapest / balanced / flagship). | | search_docs | Full-text search across every documentation page (multi-language); returns titles, URLs and snippets. | | get_doc | Fetch one documentation page as clean Markdown by path (e.g. "/api/systemone/"). | ### Client configuration Claude Code (or any MCP client): ```json { "mcpServers": { "easyjev": { "type": "http", "url": "https://api.easyjev.online/mcp" } } } ``` ## One-shot: ask an agent to onboard itself Paste this prompt into any coding assistant — it reads the docs and writes the integration for you: ```prompt Read https://docs.easyjev.online/llms.txt and integrate the EasyJev decision-model API into this project: 1. Create a client using the official typesafe-sdk with base_url = https://api.easyjev.online and an sk- key from the env var EASYJEV_API_KEY. 2. Implement one judgment call (noul) following https://docs.easyjev.online/quickstart.md 3. Handle 401/402/429 per https://docs.easyjev.online/api/errors.md ``` --- # FAQ > Frequently asked questions about EasyJev: what decision models are, typesafe-sdk compatibility, pricing and free credits, rate limits, top-up, security of sk- keys and where to get help. ## What is a decision model, and how is it different from an LLM? Decision models (the “System One” family) are discriminative: given a state and typed questions, they return structured judgments — yes/no probabilities, option distributions, ordinal scores — instead of generating text. You get a typed, parseable answer in one forward pass, without paying for generation. ## Is the official typesafe-sdk really compatible? Yes — swap `base_url` to `https://api.easyjev.online` and use an sk- key issued by this site. Request shapes, error classes and `request_id` tracing follow the official contract; multimodal assets (SDK `Image`) are supported too. ## How much does it cost? Are output tokens really free? Pay-as-you-go, per model, per million input tokens — per model, per million input tokens (check the live catalog for every current price); output tokens are free on every model currently on sale. Failed calls (4xx/5xx/timeouts) are never billed. Check [live prices](https://docs.easyjev.online/models/) or the MCP `list_models` tool. ## Do new accounts get free credit? Yes — the first email-code sign-in creates the account and lands a small free credit automatically (the exact amount is a runtime setting; the console shows the live value, and the MCP `get_free_credit_policy` tool reports it too). ## What are the rate limits? Default per key: 600 requests/minute and 4,500,000 tokens/minute (sliding window; the console shows your live values). Exceeding either returns 429 — back off briefly and retry; media bytes are excluded from the estimate. ## How do I check my balance from code? `GET /v1/balance` with the same sk- key — free, no tokens consumed. It returns the balance in quota and USD plus your low-balance alert state. See the [reference](https://docs.easyjev.online/api/balance/). ## Can AI agents read these docs directly? Yes — start from `/llms.txt`, read any page as Markdown by appending `.md`, or connect the read-only MCP server at `api.easyjev.online/mcp` (live prices, model comparison, doc search). See [Agents & MCP](https://docs.easyjev.online/agents/). ## Is my key safe? What do you store? API keys are shown once and stored as SHA-256 hashes only — a lost key is deleted and replaced, never “recovered”. Login is email-code only (no passwords; codes are stored hashed, single-use, 10-minute validity). Images sent to multimodal calls never touch the database; request archives truncate to ~64 KB with media redacted. --- # typesafe/jev-1.13 — API, pricing & docs > Call typesafe/jev-1.13 on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is typesafe/jev-1.13 Jev is the first System One decision model by TypeSafe AI. EasyJev hosts the full Jev lineup (typesafe/jev-1.13, typesafe/jev-1.14 and the typesafe/jev-latest alias) behind one typesafe-sdk-compatible endpoint. TypeSafe: Jev 1.13 · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "typesafe/jev-1.13", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [typesafe/jev-1.14](https://docs.easyjev.online/models/typesafe-jev-1-14/) - [typesafe/jev-latest](https://docs.easyjev.online/models/typesafe-jev-latest/) --- # typesafe/jev-1.14 — API, pricing & docs > Call typesafe/jev-1.14 on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is typesafe/jev-1.14 Jev is the first System One decision model by TypeSafe AI. EasyJev hosts the full Jev lineup (typesafe/jev-1.13, typesafe/jev-1.14 and the typesafe/jev-latest alias) behind one typesafe-sdk-compatible endpoint. TypeSafe: Jev 1.14 · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "typesafe/jev-1.14", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [typesafe/jev-1.13](https://docs.easyjev.online/models/typesafe-jev-1-13/) - [typesafe/jev-latest](https://docs.easyjev.online/models/typesafe-jev-latest/) --- # typesafe/jev-latest — API, pricing & docs > Call typesafe/jev-latest on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is typesafe/jev-latest Jev is the first System One decision model by TypeSafe AI. EasyJev hosts the full Jev lineup (typesafe/jev-1.13, typesafe/jev-1.14 and the typesafe/jev-latest alias) behind one typesafe-sdk-compatible endpoint. TypeSafe: Jev (latest) · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "typesafe/jev-latest", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [typesafe/jev-1.13](https://docs.easyjev.online/models/typesafe-jev-1-13/) - [typesafe/jev-1.14](https://docs.easyjev.online/models/typesafe-jev-1-14/) --- # baseten/joey-1 — API, pricing & docs > Call baseten/joey-1 on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is baseten/joey-1 Hosted on Baseten: a family of decision-model deployments (Joey, Elephant, Mammoth) for structured judgments. Baseten: Joey 1 · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "baseten/joey-1", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [baseten/elephant-1](https://docs.easyjev.online/models/baseten-elephant-1/) - [baseten/mammoth-1](https://docs.easyjev.online/models/baseten-mammoth-1/) --- # baseten/elephant-1 — API, pricing & docs > Call baseten/elephant-1 on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is baseten/elephant-1 Hosted on Baseten: a family of decision-model deployments (Joey, Elephant, Mammoth) for structured judgments. Baseten: Elephant 1 · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "baseten/elephant-1", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [baseten/joey-1](https://docs.easyjev.online/models/baseten-joey-1/) - [baseten/mammoth-1](https://docs.easyjev.online/models/baseten-mammoth-1/) --- # baseten/mammoth-1 — API, pricing & docs > Call baseten/mammoth-1 on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is baseten/mammoth-1 Hosted on Baseten: a family of decision-model deployments (Joey, Elephant, Mammoth) for structured judgments. Baseten: Mammoth 1 · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "baseten/mammoth-1", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [baseten/joey-1](https://docs.easyjev.online/models/baseten-joey-1/) - [baseten/elephant-1](https://docs.easyjev.online/models/baseten-elephant-1/) --- # cloudflare/clairic — API, pricing & docs > Call cloudflare/clairic on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is cloudflare/clairic Clef is Cloudflare’s decision-model line for edge-speed judgments; Clef, Clef Flash, Clef Omni and Clairic are served through this gateway. Cloudflare: Clairic · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "cloudflare/clairic", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [cloudflare/clef](https://docs.easyjev.online/models/cloudflare-clef/) - [cloudflare/clef-flash](https://docs.easyjev.online/models/cloudflare-clef-flash/) - [cloudflare/clef-omni](https://docs.easyjev.online/models/cloudflare-clef-omni/) --- # cloudflare/clef — API, pricing & docs > Call cloudflare/clef on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is cloudflare/clef Clef is Cloudflare’s decision-model line for edge-speed judgments; Clef, Clef Flash, Clef Omni and Clairic are served through this gateway. Cloudflare: Clef · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "cloudflare/clef", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [cloudflare/clairic](https://docs.easyjev.online/models/cloudflare-clairic/) - [cloudflare/clef-flash](https://docs.easyjev.online/models/cloudflare-clef-flash/) - [cloudflare/clef-omni](https://docs.easyjev.online/models/cloudflare-clef-omni/) --- # cloudflare/clef-flash — API, pricing & docs > Call cloudflare/clef-flash on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is cloudflare/clef-flash Clef is Cloudflare’s decision-model line for edge-speed judgments; Clef, Clef Flash, Clef Omni and Clairic are served through this gateway. Cloudflare: Clef Flash · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "cloudflare/clef-flash", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [cloudflare/clairic](https://docs.easyjev.online/models/cloudflare-clairic/) - [cloudflare/clef](https://docs.easyjev.online/models/cloudflare-clef/) - [cloudflare/clef-omni](https://docs.easyjev.online/models/cloudflare-clef-omni/) --- # cloudflare/clef-omni — API, pricing & docs > Call cloudflare/clef-omni on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is cloudflare/clef-omni Clef is Cloudflare’s decision-model line for edge-speed judgments; Clef, Clef Flash, Clef Omni and Clairic are served through this gateway. Cloudflare: Clef Omni · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "cloudflare/clef-omni", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [cloudflare/clairic](https://docs.easyjev.online/models/cloudflare-clairic/) - [cloudflare/clef](https://docs.easyjev.online/models/cloudflare-clef/) - [cloudflare/clef-flash](https://docs.easyjev.online/models/cloudflare-clef-flash/) --- # nace-ai/drex-max — API, pricing & docs > Call nace-ai/drex-max on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is nace-ai/drex-max Drex is Nace.AI’s decision-model family focused on enterprise judgment workloads; the Drex Max, v2.0 and v1.5 checkpoints are all on sale here. Nace.AI: Drex Max · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "nace-ai/drex-max", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [nace-ai/drex-v2.0](https://docs.easyjev.online/models/nace-ai-drex-v2-0/) - [nace-ai/drex-v1.5](https://docs.easyjev.online/models/nace-ai-drex-v1-5/) --- # nace-ai/drex-v2.0 — API, pricing & docs > Call nace-ai/drex-v2.0 on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is nace-ai/drex-v2.0 Drex is Nace.AI’s decision-model family focused on enterprise judgment workloads; the Drex Max, v2.0 and v1.5 checkpoints are all on sale here. Nace.AI: Drex v2.0 · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "nace-ai/drex-v2.0", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [nace-ai/drex-max](https://docs.easyjev.online/models/nace-ai-drex-max/) - [nace-ai/drex-v1.5](https://docs.easyjev.online/models/nace-ai-drex-v1-5/) --- # nace-ai/drex-v1.5 — API, pricing & docs > Call nace-ai/drex-v1.5 on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is nace-ai/drex-v1.5 Drex is Nace.AI’s decision-model family focused on enterprise judgment workloads; the Drex Max, v2.0 and v1.5 checkpoints are all on sale here. Nace.AI: Drex v1.5 · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "nace-ai/drex-v1.5", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [nace-ai/drex-max](https://docs.easyjev.online/models/nace-ai-drex-max/) - [nace-ai/drex-v2.0](https://docs.easyjev.online/models/nace-ai-drex-v2-0/) --- # liquid/d1 — API, pricing & docs > Call liquid/d1 on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is liquid/d1 d1 is Liquid AI’s decision-model line, benchmarked publicly against other System One models and served here across d1, d1.1 and d1-ultra. LiquidAI: d1 · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "liquid/d1", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [liquid/d1.1](https://docs.easyjev.online/models/liquid-d1-1/) - [liquid/d1-ultra](https://docs.easyjev.online/models/liquid-d1-ultra/) --- # liquid/d1.1 — API, pricing & docs > Call liquid/d1.1 on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is liquid/d1.1 d1 is Liquid AI’s decision-model line, benchmarked publicly against other System One models and served here across d1, d1.1 and d1-ultra. LiquidAI: d1.1 · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "liquid/d1.1", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [liquid/d1](https://docs.easyjev.online/models/liquid-d1/) - [liquid/d1-ultra](https://docs.easyjev.online/models/liquid-d1-ultra/) --- # liquid/d1-ultra — API, pricing & docs > Call liquid/d1-ultra on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is liquid/d1-ultra d1 is Liquid AI’s decision-model line, benchmarked publicly against other System One models and served here across d1, d1.1 and d1-ultra. LiquidAI: d1-ultra · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "liquid/d1-ultra", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [liquid/d1](https://docs.easyjev.online/models/liquid-d1/) - [liquid/d1.1](https://docs.easyjev.online/models/liquid-d1-1/) --- # microsoft/microsoft-decision-1 — API, pricing & docs > Call microsoft/microsoft-decision-1 on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is microsoft/microsoft-decision-1 microsoft-decision-1 is Microsoft’s entry in the System One decision-model catalog. Microsoft: Microsoft-Decision-1 · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "microsoft/microsoft-decision-1", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. --- # upstage/solar-decide — API, pricing & docs > Call upstage/solar-decide on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is upstage/solar-decide Solar Decider is Upstage’s decision model built on the Solar backbone, in standard and Flash variants. Upstage: Solar Decide · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "upstage/solar-decide", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [upstage/solar-decide-flash](https://docs.easyjev.online/models/upstage-solar-decide-flash/) --- # upstage/solar-decide-flash — API, pricing & docs > Call upstage/solar-decide-flash on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is upstage/solar-decide-flash Solar Decider is Upstage’s decision model built on the Solar backbone, in standard and Flash variants. Upstage: Solar Decide Flash · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "upstage/solar-decide-flash", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [upstage/solar-decide](https://docs.easyjev.online/models/upstage-solar-decide/) --- # morph/reflex — API, pricing & docs > Call morph/reflex on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is morph/reflex Reflex is Morph’s decision model in the System One catalog, served through this gateway. Morph: Reflex · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "morph/reflex", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. --- # parallel/decisions-fast — API, pricing & docs > Call parallel/decisions-fast on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is parallel/decisions-fast Decisions Fast is Parallel’s speed-focused decision model — the cheapest tier in this catalog. Parallel: Decisions Fast · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "parallel/decisions-fast", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. --- # perplexity/pplx-decider-v1.2 — API, pricing & docs > Call perplexity/pplx-decider-v1.2 on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is perplexity/pplx-decider-v1.2 pplx-decider is Perplexity’s family of decision models (built on a 27B backbone) with a dedicated Decisions API, free output tokens and an open-weight release on Hugging Face. Perplexity: Decider V1.2 · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "perplexity/pplx-decider-v1.2", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [perplexity/pplx-decider-flash-v1.2](https://docs.easyjev.online/models/perplexity-pplx-decider-flash-v1-2/) --- # perplexity/pplx-decider-flash-v1.2 — API, pricing & docs > Call perplexity/pplx-decider-flash-v1.2 on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is perplexity/pplx-decider-flash-v1.2 pplx-decider is Perplexity’s family of decision models (built on a 27B backbone) with a dedicated Decisions API, free output tokens and an open-weight release on Hugging Face. Perplexity: Decider Flash V1.2 · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "perplexity/pplx-decider-flash-v1.2", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [perplexity/pplx-decider-v1.2](https://docs.easyjev.online/models/perplexity-pplx-decider-v1-2/) --- # respan/span-02 — API, pricing & docs > Call respan/span-02 on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is respan/span-02 Span is Respan’s decision-model family (Span-02, Span-01 and the Lite variant). Respan: Span-02 · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "respan/span-02", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [respan/span-01](https://docs.easyjev.online/models/respan-span-01/) - [respan/span-01-lite](https://docs.easyjev.online/models/respan-span-01-lite/) --- # respan/span-01 — API, pricing & docs > Call respan/span-01 on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is respan/span-01 Span is Respan’s decision-model family (Span-02, Span-01 and the Lite variant). Respan: Span-01 · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "respan/span-01", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [respan/span-02](https://docs.easyjev.online/models/respan-span-02/) - [respan/span-01-lite](https://docs.easyjev.online/models/respan-span-01-lite/) --- # respan/span-01-lite — API, pricing & docs > Call respan/span-01-lite on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is respan/span-01-lite Span is Respan’s decision-model family (Span-02, Span-01 and the Lite variant). Respan: Span-01 Lite · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "respan/span-01-lite", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. ### Other models - [respan/span-02](https://docs.easyjev.online/models/respan-span-02/) - [respan/span-01](https://docs.easyjev.online/models/respan-span-01/) --- # scaledown/classify-1 — API, pricing & docs > Call scaledown/classify-1 on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is scaledown/classify-1 classify-1 is ScaledDown’s compact classification-focused decision model. ScaleDown: Classify-1 · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "scaledown/classify-1", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. --- # togethercomputer/tev1-4b-experimental — API, pricing & docs > Call togethercomputer/tev1-4b-experimental on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is togethercomputer/tev1-4b-experimental TEV is an experimental 4B decision model from the Together ecosystem. Together: Tev1 4B Experimental · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "togethercomputer/tev1-4b-experimental", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request. --- # jaredpalmer/kev-4b — API, pricing & docs > Call jaredpalmer/kev-4b on EasyJev with live pricing, vendor facts and a copy-paste curl example. typesafe-sdk compatible — swap base_url to https://api.easyjev.online and keep everything else. ## What is jaredpalmer/kev-4b kev-4b is a community 4B decision model published by Jared Palmer. Jared Palmer: Kev 4B · served through the EasyJev gateway behind a single typesafe-sdk-compatible endpoint — swap base_url and api_key and keep everything else. ## Pricing | Vendor | Released | Input price | Output price | | --- | --- | --- | --- | Everything below is fetched live from [GET /v1/models](https://docs.easyjev.online/models/api/models/) at request time (never a build-time snapshot); the MCP `list_models` tool serves the same data. ## Call this model One endpoint for the whole catalog — this model is selected by the `model` field: ```curl curl -X POST "https://api.easyjev.online/v1/systemone" \ -H "Authorization: Bearer sk-your-key" \ -H "Content-Type: application/json" \ -d '{ "model": "jaredpalmer/kev-4b", "state": "User asks: can I get a refund for order #1024?", "questions": { "refund": {"type": "noul", "instructions": "Is this a refund request?"} } }' ``` Switching models is a one-line change — same request shape across the entire catalog, so you can A/B test models per request.