OpenRouter API Relay

OpenRouter is an LLM gateway. One key gives access to hundreds of models (OpenAI / Anthropic / Google / Meta / Mistral, etc.) using vendor/model naming.

POST https://www.myai168.com/en/api/openrouter/v1/chat/completions

Chat Completions (OpenAI format compatible).

curl https://www.myai168.com/en/api/openrouter/v1/chat/completions \
  -H "Authorization: Bearer YOUR_DEV_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"anthropic/claude-sonnet-5","messages":[{"role":"user","content":"Hello"}]}'

import requests

r = requests.post(
    "https://www.myai168.com/en/api/openrouter/v1/chat/completions",
    headers={"Authorization": "Bearer YOUR_DEV_KEY"},
    json={"model": "anthropic/claude-sonnet-5", "messages": [{"role": "user", "content": "Hello"}]},
)
print(r.json())

POST https://www.myai168.com/en/api/openrouter/v1/chat/responses

Note: this endpoint is NOT the OpenAI Responses API — the request and response are both in Chat Completions format (same fields as completions above). It behaves the same as completions: both endpoints pre-check your credit balance and return 402 without forwarding when credits run out. For the OpenAI Responses API schema use https://www.myai168.com/en/api/openai/v1/responses instead.

curl https://www.myai168.com/en/api/openrouter/v1/chat/responses \
  -H "Authorization: Bearer YOUR_DEV_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"openai/gpt-5.6-sol","messages":[{"role":"user","content":"Hello"}]}'

import requests

r = requests.post(
    "https://www.myai168.com/en/api/openrouter/v1/chat/responses",
    headers={"Authorization": "Bearer YOUR_DEV_KEY"},
    json={"model": "openai/gpt-5.6-sol", "messages": [{"role": "user", "content": "Hello"}]},
)
print(r.json())

POST https://www.myai168.com/en/api/openrouter/v1/audio/speech

Text-to-speech (routed through OpenRouter to TTS provider). Note: OpenRouter does not offer openai/tts-1 — current TTS model ids are versioned, e.g. openai/gpt-4o-mini-tts-2025-12-15 and mistralai/voxtral-mini-tts-2603; check the official OpenRouter model list.

curl https://www.myai168.com/en/api/openrouter/v1/audio/speech \
  -H "Authorization: Bearer YOUR_DEV_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"openai/gpt-4o-mini-tts-2025-12-15","input":"Hello","voice":"nova"}' \
  --output speech.mp3

import requests

r = requests.post(
    "https://www.myai168.com/en/api/openrouter/v1/audio/speech",
    headers={"Authorization": "Bearer YOUR_DEV_KEY"},
    json={"model": "openai/gpt-4o-mini-tts-2025-12-15", "input": "Hello", "voice": "nova"},
)
with open("speech.mp3", "wb") as f:
    f.write(r.content)

POST https://www.myai168.com/en/api/openrouter/v1/audio/transcriptions

Speech-to-text (multipart upload). OpenRouter's official STT models are openai/whisper-1 / openai/whisper-large-v3 etc. (openai/gpt-4o-transcribe is not listed); this relay accepts openai/whisper-1.

curl https://www.myai168.com/en/api/openrouter/v1/audio/transcriptions \
  -H "Authorization: Bearer YOUR_DEV_KEY" \
  -F file=@audio.mp3 \
  -F model=openai/whisper-1

import requests

with open("audio.mp3", "rb") as f:
    r = requests.post(
        "https://www.myai168.com/en/api/openrouter/v1/audio/transcriptions",
        headers={"Authorization": "Bearer YOUR_DEV_KEY"},
        files={"file": f},
        data={"model": "openai/whisper-1"},
    )
print(r.json())

POST https://www.myai168.com/en/api/openrouter/v1/video/generations

Video generation (depends on supported video models). Note: vendor/some-video-model in the examples is a placeholder, not a runnable model id — the OpenRouter video endpoint is new and the available models are announced by OpenRouter; an unsupported model gets a 4xx from the upstream (never charged).

curl https://www.myai168.com/en/api/openrouter/v1/video/generations \
  -H "Authorization: Bearer YOUR_DEV_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"vendor/some-video-model","prompt":"A cat dancing"}'

import requests

r = requests.post(
    "https://www.myai168.com/en/api/openrouter/v1/video/generations",
    headers={"Authorization": "Bearer YOUR_DEV_KEY"},
    json={"model": "vendor/some-video-model", "prompt": "A cat dancing"},
)
print(r.json())

Request parameters (chat/completions and chat/responses)

Both endpoints take a Chat Completions format JSON body (chat/responses is not the OpenAI Responses API — it behaves the same as completions). Authenticate with an Authorization: Bearer YOUR_API_KEY header (never place keys in URL query strings).

Response format

OpenAI Chat Completions compatible: text in choices[0].message.content (streaming deltas in choices[0].delta.content). Billing converts the actual cost reported by OpenRouter into points; non-streaming responses carry X-Cost-Points (points charged) and X-Credits-Remaining (balance after the call) headers, and streaming responses append the same information after the stream ends as SSE comment lines : cost=N / : credits-remaining=M. You can also query your balance any time with GET https://www.myai168.com/en/api/user/get_credits.

Error codes

No points are ever charged on errors (including when the vendor returns no usage / cost info).