Model Intelligence
Compare the evidence. Choose for the task.
Explore model capabilities, benchmarks, pricing, and evaluations where data is available.
- 01IntelligenceHigher · score
- 02SpeedHigher · tok/s
- 03LatencyLower · s TTFT
- 04Input PriceLower · $/1M tokens
- 05Output PriceLower · $/1M tokens
- 06Context WindowHigher · tokens
Source: Canonical model registry at build time
Comparison
Look beyond a model name.
Consider capability, cost, latency, context, and modality alongside the available evidence. Compare models against the work you need to do.
Evidence
Understand what each result measures.
- 01
Model facts
Specifications and capabilities described for the model.
- 02
Benchmarks
Results from a defined test.
- 03
Ethen evaluations
Ethen's assessments, where published.
- 04
Runtime availability
Where a model can be accessed.
- 05
Production outcomes
Results recorded in real use, where available.
Catalog
Explore the details behind a comparison.
Model Library organizes publishers, families, variants, and metadata.
No certified ranking evidence is published yet.
- Catalog profiles
- 550
- Providers
- 53
- Categories
- 8
Access
Check where you can run it.
A comparison profile does not establish hosting or runtime access. A model may be served by different providers; execution depends on supported provider access or a compatible local runtime.
FAQ
How does Model Intelligence differ from the two libraries?
Model Intelligence owns comparisons, benchmarks, leaderboards, and evaluation evidence. The Flagship Model Library is the curated catalog; the Open Source Model Library is the separate accepted public model corpus.
Does every model have certified benchmark results?
No. Available data and unknowns remain visible. Leaderboard rankings are published only when certified results are available.
Keep exploring your options.
Discover models or review supported access through Gateway.