API reference

Base URL

https://codexkey.ru/v1

All endpoints are OpenAI-compatible — you can use the official SDK by overriding base_url.

POST /v1/chat/completions

Main endpoint for GPT-5.x chat models and gpt-5.4-mini.

Request parameters

Parameter	Type	Required	Description
model	string	yes	Model identifier (gpt-5.4, gpt-5.5, …).
messages	array	yes	Array of {role, content} messages.
temperature	number	no	0.0–2.0. Default 1.0.
top_p	number	no	Nucleus sampling, 0.0–1.0.
max_tokens	integer	no	Maximum tokens in the response.
stream	boolean	no	If true — Server-Sent Events.
reasoning_effort	string	no	minimal / low / medium / high.
service_tier	string	no	default / priority / flex.
tools	array	no	Function declarations for tool use.
tool_choice	string|obj	no	auto / none / {type: "function", ...}.

Example

curl https://codexkey.ru/v1/chat/completions \
  -H "Authorization: Bearer sk-clb-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.4",
    "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "Write hello world in Rust."}
    ],
    "temperature": 0.2,
    "max_tokens": 500
  }'

Response

{
  "id": "chatcmpl-9X...",
  "object": "chat.completion",
  "created": 1747641600,
  "model": "gpt-5.4",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "fn main() { println!(\"Hello, world!\"); }"
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 24,
    "completion_tokens": 14,
    "reasoning_tokens": 0,
    "total_tokens": 38
  }
}

POST /v1/responses

Endpoint for Codex models and agent sessions with tool use. Supports streamed events and stateful mode.

Parameters

Parameter	Type	Required	Description
model	string	yes	codex-5.3 or gpt-5.4 with Responses API support.
input	string	yes	Prompt text or array of messages.
instructions	string	no	System instructions.
tools	array	no	Functions / built-in tools.
previous_response_id	string	no	Continue a session — reference to a previous response.
stream	boolean	no	SSE event stream.

Example

curl https://codexkey.ru/v1/responses \
  -H "Authorization: Bearer sk-clb-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "codex-5.3",
    "input": "Create a file hello.py with print(\"hi\")"
  }'

Response

{
  "id": "resp_abc123",
  "object": "response",
  "created_at": 1747641600,
  "model": "codex-5.3",
  "output": [
    {
      "type": "message",
      "role": "assistant",
      "content": [{"type": "output_text", "text": "Done. File hello.py created."}]
    }
  ],
  "usage": {
    "input_tokens": 18,
    "output_tokens": 12,
    "reasoning_tokens": 4,
    "total_tokens": 34
  }
}

GET /v1/models

Returns the list of models available to the current key.

curl https://codexkey.ru/v1/models \
  -H "Authorization: Bearer sk-clb-..."

{
  "object": "list",
  "data": [
    {"id": "gpt-5.4", "object": "model", "owned_by": "codexkey"},
    {"id": "gpt-5.5", "object": "model", "owned_by": "codexkey"},
    {"id": "gpt-5.4-mini", "object": "model", "owned_by": "codexkey"},
    {"id": "codex-5.3", "object": "model", "owned_by": "codexkey"}
  ]
}

Error codes

Code	Meaning	What to do
200	OK	Success.
400	Bad Request	Check JSON and required parameters.
401	Unauthorized	Verify the Authorization header and the key.
402	Payment Required	Balance depleted — top up at /cabinet.
429	Too Many Requests	Backoff — respect Retry-After.
500	Internal Server Error	Transient error — retry with delay.
503	Service Unavailable	Upstream overload — retry in 5–30 seconds.

Error format

{
  "error": {
    "type": "insufficient_quota",
    "code": "balance_zero",
    "message": "Account balance is zero. Top up at https://codexkey.ru/cabinet/billing.",
    "param": null
  }
}

Rate limits

Default per-key limits:

• 60 RPM (requests per minute),
• 200k TPM (tokens per minute).

Limits scale automatically with spend growth. Current limits are visible at /cabinet/limits.