Chat completions

Every field we read, every limit we apply, and what comes back.

Checked against the code on 15 Sept 2026

Endpoint

POST https://heyaskr.ai/v1/chat/completions

Create a model response for a conversation. The request is the OpenAI shape. This page is precise about which fields we read, because the rest are dropped without a warning.

Request fields we read

Field	Type	What it does
model	string	A chat model id from the catalog. Exact match; there are no suffixes or aliases.
messages	array	The conversation. Each item has `role` (`system`, `user` or `assistant`) and a string `content`.
max_tokens	integer	Output ceiling, thinking included. Default 4,096, maximum 8,192. Larger values are clamped, not rejected.
stream	boolean	`true` for server-sent events. Streaming.

Everything else is dropped silently: temperature, top_p, tools, tool_choice, response_format, stop, n, seed, plugins. A request that sends them succeeds; they just have no effect. Tool calling and structured output are on the roadmap.

Messages

Roles other than system, user and assistant return 400 Unsupported message role.
content must be a string. Array content parts (image inputs) are skipped, so a message that is only parts is dropped. Image input is on the roadmap.
Only the last 40 messages are sent to the model. Trim on your side if you want control over what is kept.
More than 120,000 characters in total returns 400 The conversation is too long.
No usable messages returns 400 No usable messages were provided.

Example

curl https://heyaskr.ai/v1/chat/completions \
  -H "Authorization: Bearer $ASKR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-5",
    "max_tokens": 300,
    "messages": [
      {"role": "system", "content": "Answer in one sentence."},
      {"role": "user", "content": "Why is the sky blue?"}
    ]
  }'

Response

200

{
  "id": "chatcmpl-9c2b...",
  "object": "chat.completion",
  "created": 1789516800,
  "model": "claude-sonnet-5",
  "choices": [
    {
      "index": 0,
      "message": { "role": "assistant", "content": "Sunlight scatters off air molecules, and blue scatters most." },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 21,
    "completion_tokens": 14,
    "total_tokens": 35,
    "completion_tokens_details": { "reasoning_tokens": 0 }
  },
  "askr": { "credits_charged": 0.192 }
}

Field	Meaning
choices[0].message.content	The reply. Always one choice.
choices[0].message.images	Present only when the model answered with a picture. Images from chat.
usage.completion_tokens_details.reasoning_tokens	On a thinking model, output tokens you never see. They are billed like any other output token.
askr.credits_charged	What the call cost, in credits, to four decimal places. Additive: clients that do not know it ignore it.

How the cost is worked out

Before the call, credits are reserved for the worst case: your estimated input tokens at the model's input rate, plus max_tokens at its output rate. When the model finishes, the reservation is settled to the real cost, using the gateway's own figure for the call. Nothing else is added: there is no platform fee in the first version. Holds and settlement.

Limits

Requests: 120 per minute per IP address
Body size: 256 KB
Output: max_tokens up to 8,192; default 4,096
Deadline: 120 seconds upstream. Past it you get 504 and the reservation is charged, because the model did run. Errors