Inovacc Developer

API reference / AI chat

Create a chat completion

POSThttps://ai.inovacc.dev/v1/chat/completions

Generate the next message of a conversation, in the OpenAI chat-completions format.

Body params

  • messagesarrayrequired

    The conversation: objects with a role (system, user, assistant or tool) and a content, either a string or an array of text and image_url parts. At most 20 images.

  • modelstring

    A route id from GET /v1/models. auto, or no value, uses your organization's default route.

  • streamboolean

    true sends the reply as server-sent events, ending with data: [DONE].

  • stream_optionsobject

    With {"include_usage": true}, a stream ends with a chunk that carries the token counts.

  • max_tokensinteger

    The most output tokens you want. It can only lower the ceiling your route and organization set.

  • max_completion_tokensinteger

    Same as max_tokens, under the newer name.

  • temperaturenumber

    Sampling temperature, passed to the model.

  • top_pnumber

    Nucleus sampling, passed to the model.

  • stopstring | array

    Sequences that end the reply.

  • presence_penaltynumber

    Penalty for repeating topics, passed to the model.

  • frequency_penaltynumber

    Penalty for repeating tokens, passed to the model.

  • logit_biasobject

    Token biases, passed to the model.

  • logprobsboolean

    Ask for log probabilities of the output tokens.

  • top_logprobsinteger

    How many alternatives per token to return with logprobs.

  • toolsarray

    Functions the model may call. Some routes take a narrower set of fields and refuse it (unsupported_field_for_route).

  • tool_choicestring | object

    Whether, and which, tool the model must call.

  • parallel_tool_callsboolean

    Whether the model may call several tools at once.

  • response_formatobject

    The output format, for example {"type": "json_object"}.

  • seedinteger

    A seed for more repeatable sampling.

  • reasoning_effortstring

    How much the model should reason, on models that do.

  • verbositystring

    How long the model should answer, on models that support it.

  • userstring

    Your end user's id. Replaced, before it leaves Inovacc, by a value derived from your organization.

  • safety_identifierstring

    A stable id of your end user for safety purposes; derived from your organization like user.

  • prompt_cache_keystring

    A key that groups requests for caching; derived from your organization like user.

  • predictionobject

    Predicted output content, passed to the model.

  • ninteger

    Accepted only when your organization has been granted this field (403 field_not_allowed otherwise).

  • service_tierstring

    Accepted only when your organization has been granted this field (403 field_not_allowed otherwise).

  • storeboolean

    Accepted only when your organization has been granted this field (403 field_not_allowed otherwise).

  • metadataobject

    Accepted only when your organization has been granted this field (403 field_not_allowed otherwise).

  • modalitiesarray

    Accepted only when your organization has been granted this field (403 field_not_allowed otherwise).

  • audioobject

    Accepted only when your organization has been granted this field (403 field_not_allowed otherwise).

  • web_search_optionsobject

    Accepted only when your organization has been granted this field (403 field_not_allowed otherwise).

Headers

  • Authorizationstringrequired

    Bearer <API key>

  • X-Operation-Idstringrequired

    a unique id you choose, on every call except GET

  • Content-Typestringrequired

    application/json, on a call with a body

See also

Source details

Catalogue entry

Id
identity/component-taxonomy/capabilities#ai.chat
Repository
identity
Path
contracts/component-taxonomy/catalog/capabilities.json
Commit
08ef3cd75d82

Contract section

Id
ai/ai/contract#post-v1chatcompletions
Repository
ai
Path
apps/cloudflare/workers/api/ai/CONTRACT.md
Commit
972a1c15b9d5
Version documented
2.37.0
Version live
2.14.0
Updated 2026-10-10.