Clef

POST /v1/systemone model: Clef

Clef is Cloudflare’s 27B decision model, post-trained from Qwen3.8-27B. Use it where you would otherwise ask a chat model to classify, route or grade something and then parse its reply.

At a glance


Endpoint	`POST /v1/systemone`
Input	a `state` (text or any JSON) and optional images
Output	one answer per question, with probabilities
Question types	`choice`, `noul`, `score`
Streaming	❌
Billing	per input token, reported as `usage.input_tokens`; `output_tokens` is always `0`
State length	truncated to 16,384 tokens
Stability	`experimental`

Request

Field
`model`	required	`Clef`
`state`	required	the situation to decide about: a string, or any JSON value
`questions`	required	an object mapping your own question names to questions, at least one
`images`	optional	an array of images as `data:` URL strings

temperature, prompt_format_version and return_prompt_token_ids are refused by name with a 422. Unknown keys inside a question are refused too, so a misspelled instruction fails loudly instead of being ignored.

Question types

`type`	`criteria`	Answer
`choice`	an object of options, `{"name": "what this option means", …}`, 1–255 entries	`choice`, `confidence`, `probabilities` per option
`noul`	optional `{"true": "…", "false": "…"}`; give `instructions`, `criteria`, or both	`noul`: the probability that the answer is yes (or that `true` holds), 0–1
`score`	an ordered array of 1–10 levels, lowest first	`score` (the expected level, counting from 0), `confidence`, `legend`, `probabilities` per level

Every question can carry instructions, the question itself in plain words.

Response

{
  "answers": {
    "status": {
      "type": "choice",
      "choice": "overdue",
      "confidence": 0.9895,
      "probabilities": { "paid": 0.0041, "overdue": 0.9930, "draft": 0.0029 }
    },
    "large": { "type": "noul", "noul": 0.9934 },
    "urgency": {
      "type": "score",
      "score": 1.5326,
      "confidence": 0.2989,
      "legend": { "0": "Can wait", "1": "This week", "2": "Today" },
      "probabilities": { "0": 0.0575, "1": 0.3524, "2": 0.5901 }
    }
  },
  "usage": { "input_tokens": 340, "output_tokens": 0 },
  "model": "Clef"
}

Use the probabilities rather than only the top answer: a noul of 0.55 and one of 0.99 are both “yes”, but only one of them is safe to act on automatically.

Example

curl https://developer.amd.com.cn/radeon/api/v1/systemone \
  -H "Authorization: Bearer $RADEON_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "Clef",
    "state": {"invoice": {"vendor": "Acme", "total": 1250.0, "currency": "USD", "status": "overdue"}},
    "questions": {
      "status": {
        "type": "choice",
        "instructions": "What is the invoice status?",
        "criteria": {"paid": "Invoice is paid.", "overdue": "Invoice is past due.", "draft": "Not sent."}
      },
      "large": {"type": "noul", "instructions": "Is the total above 1000 USD?"},
      "urgency": {"type": "score", "instructions": "How urgent is follow-up?", "criteria": ["Can wait", "This week", "Today"]}
    }
  }'

With an image:

{
  "model": "Clef",
  "state": "Look at the attached picture.",
  "images": ["data:image/png;base64,iVBORw0KGgo..."],
  "questions": {
    "colour": {
      "type": "choice",
      "instructions": "What colour fills the image?",
      "criteria": {"red": "Red", "green": "Green", "blue": "Blue"}
    }
  }
}