36 models
13 providers
Updated May 2026
MMLU saturated — now featuring GPQA Diamond & SWE-bench
Best GPQA Diamond
94.3%
Gemini 3.1 Pro
Best SWE-bench
87.6%
Claude Opus 4.7
Fastest API
400 t/s
Gemini 2.5 Flash-Lite
Cheapest Input
$0.02
per 1M tokens (Qwen3.5 2B)
Largest Context
10M tok
Llama 4 Scout
Model Parameters vs MMLU Score
Log x-axis · ◆ = MoE model · † = estimated params for closed models · hover for full details
Model Parameters vs Generation Speed
Tokens/second via provider API · Source: Artificial Analysis · ◆ = MoE · Log x-axis
Model Parameters vs Input Price ($/1M tokens)
Log-log scale · Open-source prices via hosted APIs (Together, Fireworks, etc.)
Performance vs Price — Efficiency Frontier
GPQA Diamond score vs input price · Bubble size ∝ total parameters · Upper-left = best value
⬆ Upper-left = high performance + low cost. Bubble size ∝ total parameters. ◆ diamonds = MoE architectures (active params far smaller than total).
Sort by
Model Provider Type Params (B) MMLU GPQA ◆ SWE-bench AIME 2025 t/s In $/1M Out $/1M Ctx (K)