Every major model. Real numbers. Bottley's unbiased verdict on when each is the correct choice — and when it is not.
Select your primary use case. Bottley will highlight the correct model.
Scores derive from published benchmarks (MMLU, HumanEval, MATH, MT-Bench), provider-reported specs (context window, pricing), and community latency testing. Bottley weights task-specific performance over aggregate benchmarks. A model that scores 92 on coding and 61 on creative writing is not a 76 — it is a 92 for coding tasks. Cost figures reflect API list pricing as of June 2026. Consumer plan pricing noted separately. Full reviews with methodology →
What changed. When it changed. Bottley updates this when recommendations shift.
New model drops, pricing changes, and benchmark reversals. Bottley sends one email when something important changes. Not before.
No noise. Unsubscribe anytime.