API Compatibility
How Cheap Model supports OpenAI-compatible requests, Claude messages, and async multimodal tasks.
What compatibility means here
Cheap Model supports common integration styles through the same user-facing API key:
- OpenAI-compatible requests such as
/v1/chat/completions - Claude-style messages such as
/v1/messages - async image, video, audio, and music tasks such as
/v1/images/generations,/v1/videos/generations, and/v1/audios/generations
This keeps routing explicit: only registered models are forwarded, and unknown model names are rejected. Use versioned model names such as gpt-5.5, claude-sonnet-4-5-20250929, and gemini-3.1-pro-preview. Routing is pinned per registered model, not through family shortcuts such as gpt, claude, or gemini.
In practice, most migrations start with three changes:
- updating credentials
- pointing your client at Cheap Model connection settings
- choosing the models and routing behavior that fit each workload
Claude-style requests
Use your Cheap Model key in the Anthropic-style x-api-key header:
curl https://cheapmodel.org/v1/messages \
-H "x-api-key: cm_xxx" \
-H "anthropic-version: 2023-06-01" \
-H "content-type: application/json" \
-d '{
"model": "claude-sonnet-4-5-20250929",
"max_tokens": 1024,
"messages": [{"role":"user","content":"Hello"}]
}'Async multimodal tasks
Image, video, and music calls return a task ID. Query it through /v1/tasks/{task_id}:
curl https://cheapmodel.org/v1/images/generations \
-H "Authorization: Bearer cm_xxx" \
-H "content-type: application/json" \
-d '{
"model": "gpt-image-2",
"prompt": "A cinematic wide-angle shot of a futuristic city skyline",
"size": "16:9",
"resolution": "2k"
}'What usually stays the same
- one application-level integration path for multiple providers
- the broad request and response shape your client already expects
- the ability to keep product code focused on prompts, tools, and business logic
What usually changes
- model names and provider availability
- latency and price tradeoffs by workload
- provider selection rules when you require strict upstream control
- observability expectations if you want cleaner cost and usage visibility
Migration checklist
- Confirm which workload types matter first: text, image, video, audio, or functional APIs.
- Map approved providers and models for each workload.
- Decide which models must stay pinned to a specific upstream.
- Validate billing, logging, and routing expectations before you scale traffic.
Cheap Model