Vai al contenuto
AI.info

Model Tracker

FrontierMath, tiers 1–3 leaderboard

The models in the Model Tracker that Epoch AI has measured on FrontierMath, tiers 1–3, ranked by the highest score of each: the name Epoch AI gives it, the setting it ran and the day. Epoch AI, ‘Capabilities & benchmarking’, epoch.ai/benchm

In inglese

The models in the Model Tracker that Epoch AI has measured on FrontierMath, tiers 1–3, ranked by the highest score of each: the name Epoch AI gives it, the setting it ran and the day. Epoch AI, ‘Capabilities & benchmarking’, epoch.ai/benchmarks, CC BY 4.0: Epoch AI’s own runs and its Capabilities Index.

  1. 1. GPT-6.1 Sol: 93.7% (± 1.4) — gpt-6.1-sol_max, max, 29 September 2026
  2. 1. GPT-6 Astra: 93.7% (± 1.4) — gpt-6-astra_max, max, 30 August 2026
  3. 3. Claude Opus 5.5: 91.2% (± 1.7) — claude-opus-5-5_max, max, 22 September 2026
  4. 4. Claude Fable 5.1: 90.2% (± 1.8) — claude-fable-5-1_max, max, 1 September 2026
  5. 5. GPT-6 Sol: 89.8% (± 1.8) — gpt-6-sol_max, max, 22 September 2026
  6. 6. GPT-5.6 Sol: 89.1% (± 1.8) — gpt-5.6-sol_max, max, 9 July 2026
  7. 7. Claude Sonnet 5.5: 88.8% (± 1.9) — claude-sonnet-5-5_max, max, 29 September 2026
  8. 8. GPT-5.5 Pro: 87.7% (± 1.9) — gpt-5.5-pro_xhigh, xhigh, 12 June 2026
  9. 9. Claude Fable 5: 87.0% (± 2.0) — claude-fable-5_max, max, 9 June 2026
  10. 10. Gemini 4 Argon: 86.0% (± 2.1) — gemini-4-argon_high, high, 13 October 2026
  11. 10. GPT-5.6 Terra: 86.0% (± 2.1) — gpt-5.6-terra_max, max, 9 July 2026
  12. 12. Claude Opus 5: 85.6% (± 2.1) — claude-opus-5_max, max, 24 July 2026
  13. 13. GPT-5.5: 85.3% (± 2.1) — gpt-5.5_xhigh, xhigh, 11 June 2026
  14. 14. GPT-5.4 Pro: 82.5% (± 2.3) — gpt-5.4-pro-2026-03-05_xhigh, xhigh, 13 June 2026
  15. 15. Muse Wonder: 82.1% (± 2.3) — muse-wonder-1.0_max, max, 12 October 2026
  16. 15. GPT-5.6 Luna: 82.1% (± 2.3) — gpt-5.6-luna_max, max, 9 July 2026
  17. 17. Claude Opus 4.8: 80.0% (± 2.4) — claude-opus-4-8_max, max, 10 June 2026
  18. 18. GPT-6 Luna: 78.9% (± 2.4) — gpt-6-luna_max, max, 22 September 2026
  19. 19. GPT-5.4: 78.6% (± 2.4) — gpt-5.4-2026-03-05_xhigh, xhigh, 11 June 2026
  20. 20. Claude Haiku 5.5: 75.1% (± 2.6) — claude-haiku-5-5_max, max, 7 October 2026
  21. 21. Qwen3.8-Max: 74.7% (± 2.6) — qwen3.8-max_xhigh, xhigh, 4 August 2026
  22. 22. GPT-5.2 Pro: 74.0% (± 2.6) — gpt-5.2-pro-2025-12-11_xhigh, xhigh, 13 June 2026
  23. 23. Gemini 3.7 Flash: 71.6% (± 2.7) — gemini-3.7-flash_high, high, 14 August 2026
  24. 24. Claude Opus 4.7: 70.2% (± 2.7) — claude-opus-4-7_max, max, 10 June 2026
  25. 25. Gemini 3.8 Flash: 68.4% (± 2.8) — gemini-3.8-flash_high, high, 2 September 2026
  26. 26. GPT-5.2: 67.4% (± 2.8) — gpt-5.2-2025-12-11_xhigh, xhigh, 11 June 2026
  27. 27. DeepSeek-V4.1-Flash: 67.4% (± 2.8) — deepseek-v4.1-flash_max, max, 8 October 2026
  28. 28. Grok 4.6: 66.0% (± 2.8) — grok-4.6_xhigh, xhigh, 14 August 2026
  29. 28. Claude Opus 4.6: 66.0% (± 2.8) — claude-opus-4-6_max, max, 11 June 2026
  30. 30. Claude Sonnet 5: 65.6% (± 2.8) — claude-sonnet-5_max, max, 30 June 2026
  31. 31. Gemini 3.5 Flash: 62.8% (± 2.9) — gemini-3.5-flash_high, high, 10 June 2026
  32. 32. Gemini 3.1 Pro: 59.6% (± 2.9) — gemini-3.1-pro-preview, 11 June 2026
  33. 33. Gemini 3.6 Flash: 58.9% (± 2.9) — gemini-3.6-flash_high, high, 2 August 2026
  34. 34. Mistral Large 4: 58.6% (± 2.9) — mistral-large-4-0_high, high, 12 October 2026
  35. 35. Grok 4.5: 57.2% (± 2.9) — grok-4.5_high, high, 9 July 2026
  36. 36. GPT-5 Pro: 55.8% (± 2.9) — gpt-5-pro-2025-10-06_high, high, 12 June 2026
  37. 37. GPT-5: 55.4% (± 2.9) — gpt-5-2025-08-07_high, high, 10 June 2026
  38. 38. Grok 4.7: 53.0% (± 3.0) — grok-4.7_xhigh, xhigh, 22 September 2026
  39. 39. Gemini 3 Flash: 51.2% (± 3.0) — gemini-3-flash-preview, 11 June 2026
  40. 39. GPT-5.4 Mini: 51.2% (± 3.0) — gpt-5.4-mini-2026-03-17_xhigh, xhigh, 12 June 2026
  41. 41. GPT-5 Mini: 46.7% (± 3.0) — gpt-5-mini-2025-08-07_high, high, 12 June 2026
  42. 42. DeepSeek V4 Pro: 45.3% (± 3.0) — deepseek-v4-pro_max, max, 17 June 2026
  43. 43. Grok 4.20: 44.9% (± 3.0) — grok-4.20-0309-reasoning, 13 July 2026
  44. 43. GPT-5.4 nano: 44.9% (± 3.0) — gpt-5.4-nano-2026-03-17_high, high, 12 June 2026
  45. 45. o4-mini: 36.1% (± 2.9) — o4-mini-2025-04-16_high, high, 11 June 2026
  46. 46. Qwen3.6-27B: 35.1% (± 2.8) — qwen3.6-27b, 28 August 2026
  47. 47. Claude Opus 4.5: 34.4% (± 2.8) — claude-opus-4-5-20251101_32K, 32K, 11 June 2026
  48. 48. o3: 33.3% (± 2.8) — o3-2025-04-16_high, high, 27 August 2026
  49. 49. Gemini 3.1 Flash-Lite: 27.7% (± 2.7) — gemini-3.1-flash-lite_high, high, 30 August 2026
  50. 50. Gemini 3.5 Flash-Lite: 26.0% (± 2.6) — gemini-3.5-flash-lite_high, high, 2 August 2026
  51. 51. Gemini 2.5 Pro: 24.6% (± 2.6) — gemini-2.5-pro, 11 June 2026
  52. 52. Claude Sonnet 4.5: 23.9% (± 2.5) — claude-sonnet-4-5-20250929_32K, 32K, 11 June 2026
  53. 53. Qwen3.6-35B-A3B: 20.4% (± 2.4) — qwen3.6-35b-a3b_none, none, 30 August 2026
  54. 54. GPT-5 nano: 20.0% (± 2.4) — gpt-5-nano-2025-08-07_high, high, 12 June 2026
  55. 55. o3-mini: 18.6% (± 2.3) — o3-mini-2025-01-31_high, high, 11 June 2026
  56. 56. o1: 14.7% (± 2.1) — o1-2024-12-17_high, high, 28 August 2026
  57. 57. Claude Opus 4.1: 12.6% (± 2.0) — claude-opus-4-1-20250805_32K, 32K, 11 June 2026
  58. 58. GPT-4.1 mini: 6.7% (± 1.5) — gpt-4.1-mini-2025-04-14, 27 August 2026
  59. 59. GPT-4.1: 6.0% (± 1.4) — gpt-4.1-2025-04-14, 27 August 2026
  60. 60. GPT-4o mini: 0.7% (± 0.5) — gpt-4o-mini-2024-07-18, 27 August 2026
  61. 60. GPT-4 Turbo: 0.7% (± 0.5) — gpt-4-turbo-2024-04-09, 27 August 2026
  62. 62. GPT-4o: 0.4% (± 0.4) — gpt-4o-2024-08-06, 27 August 2026
  63. 63. GPT-3.5 Turbo: 0.0% (± 0.0) — gpt-3.5-turbo-0125, 27 August 2026

Epoch AI: the benchmark and its data · CC BY 4.0

Other benchmarks

All of the Model Tracker