Build with Cheaper Inference

Quick start

1. Create an account.
2. Add at least $5 to your wallet. Your first successful funding adds $10 in bonus credit.
3. Create an API key from the dashboard and copy it once.
4. Set your OpenAI SDK baseURL to the endpoint below.
5. Put a supported model name in the model field.
6. Send your first request and review usage in History.

Using an AI app?

Use the step-by-step Codex CLI and desktop app guide for the Responses API, or browse the other app integrations. Each guide selects the compatible API mode and explains how to verify the connection.

cURL chat completion

curl https://api.cheaperinference.com/v1/chat/completions \
  -H "Authorization: Bearer ir_live_YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-opus-4.6",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

cURL Responses request

curl https://api.cheaperinference.com/v1/responses \
  -H "Authorization: Bearer ir_live_YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.4",
    "input": "Hello!",
    "store": false,
    "stream": true
  }'

The stateless Responses compatibility layer supports Codex streaming, function calls, and custom local tools such as apply_patch. Requests must use store: false. Stored responses, previous_response_id, conversations, background mode, and provider-hosted web or file search are not supported.

Kimi K3 reasoning effort

Use reasoning.effort to control how much reasoning Kimi K3 performs. Kimi K3 currently supports low, high, and max. medium is not a native effort level for this model. Set the effort explicitly when consistent behavior matters.

curl https://api.cheaperinference.com/v1/chat/completions \
  -H "Authorization: Bearer ir_live_YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kimi-k3",
    "messages": [{"role": "user", "content": "Solve this carefully."}],
    "reasoning": {"effort": "high"}
  }'

Lower effort is faster and uses fewer output tokens; higher effort gives the model more room for difficult work. Reasoning tokens count as output usage and are billed at the model's output-token rate.

Image generation

curl https://api.cheaperinference.com/v1/images/generations \
  -H "Authorization: Bearer ir_live_YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "nano-banana-pro",
    "prompt": "A clean product render of a banana invoice",
    "n": 1,
    "size": "1024x1024",
    "resolution": "1K",
    "quality": "high",
    "response_format": "url"
  }'

Use /v1/images/generations for every model whose /v1/models entry has type: "image" or capabilities.image_generation: true, including grok-imagine. Image-generation models are not accepted by /v1/responses, /v1/chat/completions, or /v1/completions.

Example: Grok Imagine

curl https://api.cheaperinference.com/v1/images/generations \
  -H "Authorization: Bearer ir_live_YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "grok-imagine",
    "prompt": "A cinematic city at night",
    "n": 1,
    "response_format": "url"
  }'

Nano Banana image output is billed using fixed output-token amounts for the selected resolution. Nano Banana 2 uses 747 tokens at 0.5K, 1,120 at 1K, 1,680 at 2K, and 2,520 at 4K. Nano Banana Pro uses 1,120 tokens at 1K or 2K and 2,000 at 4K.

Runnable vision request

Use an HTTPS image URL or a base64 data URL in an OpenAI image_url content part:

curl https://api.cheaperinference.com/v1/chat/completions \
  -H "Authorization: Bearer ir_live_YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.4",
    "messages": [{
      "role": "user",
      "content": [
        {"type": "text", "text": "Describe this image."},
        {
          "type": "image_url",
          "image_url": {"url": "https://example.com/image.png"}
        }
      ]
    }]
  }'

Prompt caching

cache_control is passed through when supported by the selected model. Cache reads and writes use imported model-specific rates when available. If no cache-read rate is available, cached reads use 10% of the normal input rate; cache writes without a separate rate use the normal input rate.

Temporary vision uploads

For larger base64 payloads, upload each image separately. The private file ID expires after one hour and keeps the forwarded JSON body small. Copy the returned message_content object into the request, then delete the upload early when it is no longer needed.

curl https://api.cheaperinference.com/v1/uploads \
  -H "Authorization: Bearer ir_live_YOUR_API_KEY" \
  -F "file=@tile.png"

curl https://api.cheaperinference.com/v1/chat/completions \
  -H "Authorization: Bearer ir_live_YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.4",
    "messages": [{
      "role": "user",
      "content": [
        {"type": "text", "text": "What is shown here?"},
        {
          "type": "image_file",
          "image_file": {"file_id": "img_RETURNED_ID"}
        }
      ]
    }]
  }'

curl -X DELETE \
  https://api.cheaperinference.com/v1/uploads/img_RETURNED_ID \
  -H "Authorization: Bearer ir_live_YOUR_API_KEY"

How pricing and routing work

For every request, Cheaper Inference ranks the eligible provider routes by estimated cost and starts with the lowest-cost route available for that model and request. If that provider is unavailable, has insufficient capacity, or fails before output begins, the gateway automatically tries the next-lowest-cost eligible route. Your public model ID and API response format remain unchanged.

A fallback route can have a different rate from the first route. The final customer charge follows the route that successfully serves the request, but it will never exceed the model maker's applicable direct API list price. Cheaper Inference adds no separate routing surcharge; the settled API rate is all-in.

Provider prices can change over time. Review the pricing and savings history across models, or open an individual model page for its hourly observations over the last 24 hours or 30 days. The exact amount charged for each completed request is available in your dashboard History.

Model catalog API

Use your API key to retrieve currently available models and their current catalog rates. A fallback route can change the final rate for a request, subject to the direct-list-price ceiling above. Filter by type, vision, reasoning, streaming, or provider to find models that meet a workload's requirements.

curl 'https://api.cheaperinference.com/v1/models?type=text&vision=true&streaming=true' \
  -H 'Authorization: Bearer ir_live_YOUR_API_KEY'

Token rates are USD per 1 million tokens. Fixed media rates use media_input_unit_price and media_unit_price per unit. For tiered models, use pricing.above_threshold when the request exceeds input_token_price_threshold.

The catalog identifies modality, endpoint, vision, reasoning, streaming, and image generation support. Tool calling, structured output, context limits, and accepted optional parameters can still vary by model and serving provider; verify the exact request shape before production use.

Production checklist

• Load your API key from a server-side environment variable.
• Use GET /v1/models to confirm the model is available and supports the required modality.
• Set an explicit client timeout appropriate for long model generations.
• For direct HTTP clients, retry transport errors, 429, and 5xx responses with exponential backoff.
• Do not retry 400, 401, 402, or 413 without correcting the request. For 402, add funds before retrying.
• Treat an interrupted stream as incomplete and retry the entire request.
• Review token usage, spend, and failures in History before increasing traffic.

For the generated public request, response, authentication, and error schemas, open the API reference. Check Status before escalating an availability issue.