Skip to main content
POST
/
v1
/
chat
/
completions
curl --request POST \
  --url https://api.octen.ai/v1/chat/completions \
  --header 'Content-Type: application/json' \
  --header 'x-api-key: <api-key>' \
  --data '
{
  "model": "anthropic/claude-opus-4.8",
  "messages": [
    {
      "role": "user",
      "content": "Explain attention in one sentence."
    }
  ],
  "stream": false,
  "max_tokens": 2048,
  "temperature": 1
}
'
{
  "id": "gen-1749812456-xyz7890",
  "object": "chat.completion",
  "created": 1749812456,
  "model": "anthropic/claude-opus-4.8",
  "choices": [
    {
      "index": 0,
      "finish_reason": "stop",
      "message": {
        "role": "assistant",
        "content": "Attention lets a model dynamically weight its inputs and focus on the most relevant information.",
        "refusal": null,
        "reasoning": null
      }
    }
  ],
  "usage": {
    "prompt_tokens": 18,
    "completion_tokens": 32,
    "total_tokens": 50
  }
}

Authorizations

​
x-api-key
string
header
required

API key used for request authentication. Obtain an API key before using the API. Note: A payment method is required to use the API.

Body

application/json

Request body for the Chat Completions API. Some parameters apply only to certain models; unsupported parameters are ignored for the selected model.

​
model
enum<string>
required

The model to use for chat completion.

Available options:
anthropic/claude-opus-4.8,
anthropic/claude-opus-4.6,
anthropic/claude-sonnet-4.6,
anthropic/claude-haiku-4.5,
google/gemini-3.5-flash,
google/gemini-3.1-pro-preview,
google/gemini-3.1-flash-lite,
google/gemini-3-flash-preview,
openai/gpt-5.5-pro,
openai/gpt-5.5,
openai/gpt-5.4,
moonshotai/kimi-k2.6,
moonshotai/kimi-k2.5,
minimax/minimax-m2.5,
qwen/qwen3.6-plus
​
messages
object[]
required

The conversation so far. System prompt plus user and assistant messages in chronological order.

A system prompt message that sets the behavior or context for the model.

​
tools
object[]

Tool definitions. Supports custom function tools and the built-in octen_search server tool.

​
tool_choice

Controls tool invocation. none: never call tools; auto: model decides (default); required: must call a tool. Can also be an object to force a specific tool. Only valid when tools is set.

Available options:
none,
auto,
required
​
parallel_tool_calls
boolean
default:true

Whether the model may issue multiple tool calls in one reply. When false, at most one tool call per turn.

​
stream
boolean
default:false

Whether to enable streaming output. When true, returns chat.completion.chunk objects incrementally.

​
max_tokens
integer

Maximum number of tokens the model can output. If not set, the model's internal default limit is used.

Required range: x >= 1
​
max_completion_tokens
integer

Maximum completion tokens, including reasoning and visible output tokens. If not set, the model's internal default limit is used.

Required range: x >= 1
​
temperature
number
default:1

Controls randomness in generation. Higher values produce more diverse output; lower values produce more deterministic output.

Required range: 0 <= x <= 2
​
top_p
number
default:1

Nucleus sampling. Only tokens with cumulative probability up to top_p are considered.

Required range: x <= 1
​
top_k
integer
default:0

Sample only from the top K most probable tokens. 0 disables it.

Required range: x >= 0
​
min_p
number
default:0

Minimum probability threshold relative to the most probable token. Tokens below it are filtered out. 0 disables it.

Required range: 0 <= x <= 1
​
top_a
number
default:0

Dynamic filtering threshold based on the most probable token. 0 disables it.

Required range: 0 <= x <= 1
​
repetition_penalty
number
default:1

Penalizes tokens already present in the input. Above 1 suppresses repetition; below 1 encourages it.

Required range: x <= 2
​
frequency_penalty
number
default:0

Penalizes tokens by their frequency in the output so far. Positive values reduce repetition.

Required range: -2 <= x <= 2
​
presence_penalty
number
default:0

Penalizes tokens that have already appeared. Positive values encourage new topics.

Required range: -2 <= x <= 2
​
response_format
object

Controls the output format. Some models may not support structured output and will automatically fall back to text.

​
stop
string[]

Stop sequences. Generation stops when any of these strings is encountered.

​
seed
integer

Seed for reproducibility. With the same parameters and model version, output should be as consistent as possible.

​
reasoning
object

Options for reasoning models. Sets the thinking effort and budget.

​
verbosity
enum<string>
default:medium

Controls how verbose the reply is.

Available options:
low,
medium,
high
​
logit_bias
object

A JSON object mapping token IDs to bias values (-100 to 100), added to the logits before sampling.

​
logprobs
boolean
default:false

Whether to return the log probabilities of the output tokens.

​
top_logprobs
integer

Number of most likely tokens to return at each position. Requires logprobs to be true.

Required range: 0 <= x <= 20
​
user
string

A unique identifier for the end user. Use hashed or pseudonymous identifiers to avoid passing personally identifiable information.

​
previous_response_id
string

The id of a previous response, used to chain state across turns. Only effective for Responses-API models; the gen--prefixed ids of regular completions cannot be used.

Response

Successful chat completion. When stream=false, returns a single chat.completion object. When stream=true, returns a stream of chat.completion.chunk objects (search_done, content, finish, usage), followed by data: [DONE].

A non-streaming chat completion response. Returned when stream=false.

​
id
string
required

The unique identifier for this request.

​
object
enum<string>
required

The object type, always chat.completion for non-streaming responses.

Available options:
chat.completion
​
created
integer
required

Unix timestamp (in seconds) of when the completion was created.

​
model
string
required

The model used for this completion.

​
choices
object[]
required

A list of completion choices.

​
search_results
object[]

Search results. Present only when octen_search was actually triggered.

​
usage
object

Token usage information. When stream=true, returned only in the final usage chunk.

​
warning
string

Warning message, if any.