AI models

OpenAI SDKs:     https://gw.pikeway.ai/ai/v1
Anthropic SDKs:  https://gw.pikeway.ai/ai

Authentication

Authorization: Bearer pk-… (OpenAI SDKs send this when you pass the key as apiKey) or X-API-Key: pk-…. For Claude Code set ANTHROPIC_BASE_URL=https://gw.pikeway.ai/ai and ANTHROPIC_AUTH_TOKEN=pk-…. Your balance must be above 0.

Billing

The gateway reads token usage from the model's response (or from the final events of a stream) and charges input × input price + cached input × cache price + output × output price, all per million tokens. Responses with status 400 or above are not charged. Prices per model are in GET /ai/v1/models.

Chat completions

OpenAI Chat Completions format. The body is passed to the model provider as-is, except that model is mapped to the provider's model id, and streaming requests get stream_options.include_usage = true so the gateway can meter the stream.

curl https://gw.pikeway.ai/ai/v1/chat/completions \
  -H "Authorization: Bearer $PIKEWAY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5",
    "messages": [{"role": "user", "content": "Summarize ERC-4337 in one sentence."}]
  }'

The response is the provider's OpenAI-format response, unchanged:

{
  "id": "chatcmpl-…",
  "object": "chat.completion",
  "model": "openai/gpt-5",
  "choices": [{
    "index": 0,
    "message": { "role": "assistant", "content": "ERC-4337 adds smart-contract wallets to Ethereum without protocol changes…" },
    "finish_reason": "stop"
  }],
  "usage": { "prompt_tokens": 17, "completion_tokens": 31, "total_tokens": 48 }
}

Messages (Anthropic)

Anthropic Messages format, for Claude Code and the Anthropic SDKs. anthropic-version and anthropic-beta headers are forwarded. Usage counts input_tokens plus cache reads and cache writes as input; cache reads are billed at the model's cache price.

curl https://gw.pikeway.ai/ai/v1/messages \
  -H "Authorization: Bearer $PIKEWAY_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{"model": "claude-sonnet-5", "max_tokens": 512,
       "messages": [{"role": "user", "content": "Hello"}]}'

export ANTHROPIC_BASE_URL=https://gw.pikeway.ai/ai
export ANTHROPIC_AUTH_TOKEN=pk-...
claude

Count tokens

Anthropic token counting, forwarded to the model provider. Requires a valid key and is free. Returns the provider's response.

Responses (OpenAI)

OpenAI Responses format. Usage is read from usage (or the response.completed event when streaming); input_tokens_details.cached_tokens are billed at the cache price.

curl https://gw.pikeway.ai/ai/v1/responses \
  -H "Authorization: Bearer $PIKEWAY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "gpt-5", "input": "Write a haiku about gas fees."}'

List models

All enabled models with prices and capabilities. No key needed.

curl https://gw.pikeway.ai/ai/v1/models

{
  "object": "list",
  "data": [
    {
      "id": "claude-sonnet-5",
      "object": "model",
      "owned_by": "anthropic",
      "display_name": "Claude Sonnet 5",
      "context_length": 1000000,
      "pricing": { "input_per_mtok": 2, "output_per_mtok": 10 },
      "capabilities": { "tools": true, "streaming": true, "thinking": true }
    }
  ]
}

Errors

• 400 invalid JSON body · missing 'model'
• 401 missing API key (Authorization: Bearer pk-...) · invalid or revoked API key
• 402 insufficient balance (balance is 0 or less)
• 403 key IP / origin restriction
• 404 unknown model: …
• 405 POST only
• 502 upstream LLM request failed: …
• Errors from the model provider are returned with their original status and body, and are not charged.