AI models

OpenRouter

One OpenAI-compatible gateway for many model providers.

Good for agents that need model fallback or broad model selection through one API, but provider variability needs monitoring.

Agent access evidence

API

native

OpenAI-compatible chat completions API supports many models.

Evidence

Browser

strong

Dashboard supports keys, credits, activity, and model selection.

Evidence

Account creation

strong

Self-serve signup and credit-based usage.

Evidence

Pricing clarity

native

Model pages expose provider and per-token pricing.

Evidence

Docs quality

strong

Docs explain routing, headers, parameters, and provider behavior.

Evidence

Readiness score

Usable

Usable: 65/100. Strongest in agent readability; weakest in agent safety.

Machine operability14/25

Agent safety10.4/25

Agent readability18.4/20

Auth and setup12/15

Production reliability10/15

Best for

Model routing, fallback experiments, quick access to many providers, and cost comparison.

Pricing

Public model catalog with per-token prices and provider routing details.

Auth

API keys and optional app attribution headers.

Account creation

Self-serve signup and credit-based usage.

Browser support

Dashboard supports keys, credits, activity, and model selection.

API

https://openrouter.ai/api/v1

Limitations

Key limitations: Provider-specific reliability, data policy, and model versioning need review before sensitive workloads. Model behavior, costs, rate limits, data retention, and safety policies can change by model and account tier; use evals, spend caps, and data controls for sensitive workloads. API automation needs client-side throttling, idempotency where supported, pagination limits, retries, and audit logging. Store credentials outside source control and use least-privilege scopes, short-lived tokens, or restricted keys where the product supports them.