Skip to content
AnyRouter

Public data

Benchmark scores

Public benchmarks from Epoch AI, Artificial Analysis, Aider and MaxxBench, set against AnyRouter price and measured speed, each chart with its source and fetch date.

Browse modelsTry in playground

Compare models

Pick a source and a benchmark, then 2–6 models. Only scores with a recorded source are shown.

1Gemini 4 Argon2GPT-6 Astra3Claude Sonnet 5.5

GPQA Diamond

Top 15 of 128 models, higher is better.

  1. 1Gemini 4 Argon96.4%
  2. 2GPT-6 Astra95.8%
  3. 3Claude Sonnet 5.595.6%
  4. 4Gemini 3.8 Flash95.4%
  5. 5GPT-6.1 Sol95.4%
  6. 6Gemini 3.7 Flash94.8%
  7. 7GPT-6 Sol94.3%
  8. 8GPT-6.1 Terra94.3%
  9. 9Gemini 3.1 Pro94.1%
  10. 10Grok 4.694.0%
  11. 11Claude Opus 5.594.0%
  12. 12Claude Opus 593.7%
  13. 13GPT-5.593.5%
  14. 14Kimi K393.5%
  15. 15DeepSeek V4.1 Flash 101793.2%

Source: Epoch AI (CC-BY 4.0), fetched 2026-11-09; Artificial Analysis, fetched 2026-07-30; Vendor model card

Side by side

Benchmarks at least two picked models report. Best per row in bold; — means no published score.

BenchmarkGemini 4 ArgonGPT-6 AstraClaude Sonnet 5.5
GPQA Diamond96.4%95.8%95.6%
ARC-AGI-291.3%95.0%83.9%
SciCode61.8%56.5%61.0%
OTIS Mock AIME100.0%100.0%100.0%
CursorBench52.9%—55.5%
SimpleQA (verified)76.0%75.6%46.5%

Source: Epoch AI (CC-BY 4.0), fetched 2026-11-09

GPQA Diamond vs price

Blended AnyRouter list price per 1M tokens (3:1 input:output), log scale.

Source: Epoch AI (CC-BY 4.0), fetched 2026-11-09; Artificial Analysis, fetched 2026-07-30; Vendor model card. Price: AnyRouter catalog; 8 free or unpriced models not plotted

GPQA Diamond vs speed

Output tokens per second measured on AnyRouter traffic.

No models have both values yet.

Source: Epoch AI (CC-BY 4.0), fetched 2026-11-09; Artificial Analysis, fetched 2026-07-30; Vendor model card. Speed: AnyRouter network

Filter by Category

Top Models Overview

Top 30 by Intelligence Index

Artificial Analysis benchmark scores

  1. 1Grok 4.661
  2. 2Claude Fable 559.9
  3. 3Claude Opus 558.9
  4. 4GPT-5.6 Sol57.7
  5. 5Kimi K357.1
  6. 6Claude Opus 4.855.7
  7. 7GPT-5.554.8
  8. 8Grok 4.553.8
  9. 9Claude Opus 4.753.5
  10. 10GPT-5.6 Terra51.6
  11. 11GPT-5.451.4
  12. 12GLM-5.251.1
  13. 13Muse Spark 1.150.6
  14. 14Gemini 3.5 Flash50.2
  15. 15Gemini 3.6 Flash50.1
  16. 16Gemini 3.1 Pro46.5
  17. 17GPT-5.1 Codex Max46
  18. 18Qwen3.7 Max46
  19. 19MiniMax M344.4
  20. 20GPT-5.3 Codex44.3
  21. 21Kimi K2.644.2
  22. 22Kimi K2.7 Code41.9
  23. 23Inkling40.7
  24. 24GPT-5.2 Codex40.1
  25. 25Qwen3.6 Plus39.6
  26. 26GLM-539.5
  27. 27Qwen3.7 Plus39
  28. 28Agnes 2.5 Pro38.8
  29. 29Agnes 2.5 Pro Alpha38.8
  30. 30GPT-5.6 Luna38.1

reasoning Comparison

Average across 7 benchmarks

  1. 1Gemini 3.1 Pro73.3%
  2. 2GPT-5.6 Sol70.1%
  3. 3Gemini 3.5 Flash69.5%
  4. 4Claude Fable 569.4%
  5. 5GPT-5.569.2%
  6. 6GPT-5.168.4%
  7. 7GPT-5 Codex67.5%
  8. 8Qwen3.7 Max67.4%
  9. 9GPT-5.1 Codex66.3%
  10. 10GPT-566.3%
  11. 11MiniMax M2.563.0%
  12. 12GPT-5 Mini62.2%
  13. 13GPT-5.1 Codex Mini62.0%
  14. 14o4-mini62.0%
  15. 15Kimi K2.561.9%

math Comparison

Average across 4 benchmarks

  1. 1o4-mini96.4%
  2. 2GPT-595.4%
  3. 3Gemini 2.5 Flash90.2%
  4. 4Gemini 2.5 Pro89.9%
  5. 5o184.7%
  6. 6Gemini 2.5 Flash-Lite83.6%
  7. 7Qwen3-14B79.6%
  8. 8DeepSeek-R177.8%
  9. 9o1-mini77.4%
  10. 10GPT-OSS 20B77.3%
  11. 11Gemini 3 Flash74.2%
  12. 12Qwen3-8B73.7%
  13. 13Kimi K2.571.2%
  14. 14GPT-4.1 Mini67.8%
  15. 15GPT-4.167.5%

Model Performance

Benchmark score radar chart · 1 selected