OpenPaths is an open source model router and AI gateway. One API key and one base URL (https://openpaths.io/v1) reach language, image, video, speech, music, embedding and 3D models from many providers.

Is OpenPaths compatible with the OpenAI and Anthropic SDKs?

Yes. POST /v1/chat/completions follows the OpenAI schema and POST /v1/messages follows the Anthropic schema. Change the base URL and keep your existing code.

How does OpenPaths pricing work?

Pay as you go at the provider price with near-zero markup. Requests on your own provider keys (BYOK) are recorded at $0. See https://openpaths.io/pricing.

What does openpaths/auto do?

openpaths/auto routes each request to a suitable model automatically, with automatic fallbacks when a provider fails or rate-limits.

Is there an OpenAPI spec or MCP server?

Yes. The OpenAPI 3.1 spec is at https://openpaths.io/openapi.json, the MCP endpoint is https://openpaths.io/mcp, and llms.txt plus llms-full.txt list every model id.

How do the price tiers differ between Claude Opus 5.5 and GPT-5.6 Sol?

Claude Opus 5.5: Cached input tokens bill at $0.20 per 1M (95% off the input rate). Claude Opus 5.5: Anthropic prompt caching also charges a write premium (1.25x input for 5-minute, 2x for 1-hour caches) on the first write; the calculator excludes it. GPT-5.6 Sol: No cached-input discount is configured for this model, so cache hit rate does not change its cost here.

Which is cheaper for long documents, Claude Opus 5.5 or GPT-5.6 Sol?

For 150K-token documents, Claude Opus 5.5 costs $3300.00 and GPT-5.6 Sol costs $3300.00 per 5,000 requests.

Can I call both Claude Opus 5.5 and GPT-5.6 Sol with one API key?

Yes. OpenPaths exposes both through one OpenAI-compatible API with pay-as-you-go billing at the published per-token rates; set the model to claude-opus-5-5 or gpt-5.6-sol.

Claude Opus 5.5 vs GPT-5.6 Sol Pricing (2026): Tiers, Cost per 1M Tokens

Claude Opus 5.5 is cheaper in 6 of 10 workloads. Claude Opus 5.5: $4.00 in / $20.00 out per 1M tokens. GPT-5.6 Sol: $4.00 in / $20.00 out per 1M tokens. Tier-aware cost for 10 workloads plus Artificial Analysis benchmarks.