Models
The API uses CRX model names. Names stay stable while the underlying model improves. The current generation is CRX-4; the CRX-3.5 line remains available as previous-generation. Prices are per million tokens (input / output).
| Model | Best for | Price (in / out per 1M) |
|---|---|---|
crx-4-fast | Fast, capable text + vision — best default for most apps | $3 / $15 |
crx-4-think | Extended reasoning (adaptive, high effort) | $3 / $15 |
crx-4-pro | Most capable — hard reasoning & agentic work | $16 / $76 |
crx-3-5-fast | Previous-gen fast model | $3 / $15 |
crx-3-5-think | Previous-gen extended reasoning | $3 / $15 |
crx-3-5-pro | Previous-gen most capable | $15 / $75 |
crx-3-2-fast | Older, lower-cost fast model | $3 / $15 |
crx-live | Lowest latency/cost — classification, live UX | $1 / $5 |
crx-3-2-diffusion | Text-to-image generation | ~$0.04 / image |
All chat models support text, vision, tool use, and streaming. GET /api/v1/models is the live source of truth for capabilities and pricing.
Free tokens & credits
- Every developer account includes 500,000 free tokens per month, resetting at the start of each calendar month.
- Opting into Data Sharing adds 100,000 tokens/month (600,000 total while enabled).
- Both input and output tokens count; more capable models draw down the allotment faster.
- After free tokens are used, usage bills against a prepaid credit balance. With no free tokens and no credits, requests return HTTP 402.
- Credits are purchased up front via hosted checkout on the Credit Reload page.
Rate limit
60 requests per minute per API key. Exceeding it returns HTTP 429 with a Retry-After header.
Common status codes
| Code | Meaning |
|---|---|
| 400 | Invalid request — malformed body or bad parameters |
| 401 | Authentication — missing, invalid, or revoked key |
| 402 | Insufficient credits — free tokens used up and no balance |
| 429 | Rate limited — see Retry-After |