Skip to main content
Follow DuckWeights on X

โ† Leaderboards

Leaderboard changes

One entry per day the board moved. Quiet days are omitted. Subscribe via Atom.

  1. 2026-10-17

    1 new model, 1 score move, 15 price changes, open gap 4.4

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Opus 5.5 57.6 ยท gap 4.4

    • New: Beam 1.0 (30.3)
    • Solar Pro 4 Intelligence 28.2 โ†’ 26.6
    • Gemma 4 26B A4B Blended $/M $0.17 โ†’ $0.17
    • Gemma 4 26B A4B Out $/M $0.40 โ†’ $0.37
    • KAT Coder Pro V2 Blended $/M $0.53 โ†’ $0.00
    • KAT Coder Pro V2 In $/M $0.30 โ†’ $0.00
    • KAT Coder Pro V2 Out $/M $1.20 โ†’ $0.00
    • Nemotron 3 Ultra 550B A55B Blended $/M $1.07 โ†’ $1.13
    • Nemotron 3 Ultra 550B A55B Out $/M $2.50 โ†’ $2.70
    • Nemotron 3.5 Super VL Blended $/M $0.29 โ†’ $0.31
    • Nemotron 3.5 Super VL In $/M $0.20 โ†’ $0.23
    • Qwen3 Coder Next Blended $/M $0.68 โ†’ $0.48
    • Qwen3 Coder Next In $/M $0.50 โ†’ $0.31
    • Qwen3 Coder Next Out $/M $1.20 โ†’ $1.00
    • Sarvam 30B Blended $/M $0.05 โ†’ $0.00
    • Sarvam 30B In $/M $0.03 โ†’ $0.00
    • Sarvam 30B Out $/M $0.11 โ†’ $0.00

  2. 2026-10-16

    3 new models, 2 top-10 changes, open gap 4.4

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Opus 5.5 57.6 ยท gap 4.4

    • Muse Spark Open entered open top 10 on Intelligence (#6)
    • Nex-N2-Pro (Based on Qwen3.5-397B-A17B) left open top 10 on Intelligence

  3. 2026-10-15

    2 score moves, open gap 4.4

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Opus 5.5 57.6 ยท gap 4.4

    • MiMo-V2.5 Intelligence 25.2 โ†’ 21.8
    • Mercury 2 Intelligence 13.8 โ†’ 10.8

  4. 2026-10-14

    3 price changes, open gap 4.4

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Opus 5.5 57.6 ยท gap 4.4

    • Muse Glimmer Blended $/M $0.58 โ†’ $0.53
    • Muse Glimmer In $/M $0.33 โ†’ $0.30
    • Muse Glimmer Out $/M $1.35 โ†’ $1.20

  5. 2026-10-13

    2 new models, 18 price changes, open gap 4.4

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Opus 5.5 57.6 ยท gap 4.4

    • New: Muse Wonder Full (55.8)
    • New: Muse Wonder (52.4)
    • DeepSeek V4 Flash 0420 Blended $/M $0.17 โ†’ $0.14
    • DeepSeek V4 Flash 0420 In $/M $0.13 โ†’ $0.11
    • DeepSeek V4 Flash 0420 Out $/M $0.28 โ†’ $0.23
    • Ling-3.0-flash-VL Blended $/M $0.11 โ†’ $0.00
    • Ling-3.0-flash-VL In $/M $0.07 โ†’ $0.00
    • Ling-3.0-flash-VL Out $/M $0.22 โ†’ $0.00
    • Phi-4 Mini Instruct Blended $/M $0.00 โ†’ $0.13
    • Phi-4 Mini Instruct In $/M $0.00 โ†’ $0.07
    • Phi-4 Mini Instruct Out $/M $0.00 โ†’ $0.30
    • Phi-4 Multimodal Instruct Blended $/M $0.00 โ†’ $0.14
    • Phi-4 Multimodal Instruct In $/M $0.00 โ†’ $0.08
    • Phi-4 Multimodal Instruct Out $/M $0.00 โ†’ $0.32
    • Qwen3 VL 32B Blended $/M $0.28 โ†’ $0.00
    • Qwen3 VL 32B In $/M $0.16 โ†’ $0.00
    • Qwen3 VL 32B Out $/M $0.64 โ†’ $0.00
    • Qwen3.6 Max Preview Blended $/M $2.92 โ†’ $0.00
    • Qwen3.6 Max Preview In $/M $1.30 โ†’ $0.00
    • Qwen3.6 Max Preview Out $/M $7.80 โ†’ $0.00

  6. 2026-10-10

    1 new model, 10 price changes, open gap 4.4

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Opus 5.5 57.6 ยท gap 4.4

    • Gemma 4 26B A4B Blended $/M $0.17 โ†’ $0.17
    • Gemma 4 26B A4B Out $/M $0.37 โ†’ $0.40
    • LongCat 2.0 Blended $/M $0.53 โ†’ $1.30
    • LongCat 2.0 In $/M $0.30 โ†’ $0.75
    • LongCat 2.0 Out $/M $1.20 โ†’ $2.95
    • Qwen3 Coder Next Blended $/M $0.56 โ†’ $0.68
    • Qwen3 Coder Next In $/M $0.35 โ†’ $0.50
    • QwQ 32B Blended $/M $0.74 โ†’ $0.00
    • QwQ 32B In $/M $0.66 โ†’ $0.00
    • QwQ 32B Out $/M $1.00 โ†’ $0.00

  7. 2026-10-09

    2 new models, 1 score move, open gap 4.4

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Opus 5.5 57.6 ยท gap 4.4

    • Granite 4.2 30B Intelligence 14.8 โ†’ 12.7

  8. 2026-10-07

    1 new model, 3 price changes, open gap 4.4

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Opus 5.5 57.6 ยท gap 4.4

    • New: Mistral Large 4 Preview (38.4)
    • Nemotron 3 Nano Omni 30B A3B Reasoning Blended $/M $0.42 โ†’ $0.45
    • Nemotron 3 Nano Omni 30B A3B Reasoning In $/M $0.20 โ†’ $0.30
    • Nemotron 3 Nano Omni 30B A3B Reasoning Out $/M $1.09 โ†’ $0.90

  9. 2026-10-06

    1 score move, open gap 4.4

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Opus 5.5 57.6 ยท gap 4.4

    • A.X-K2 Intelligence 22.7 โ†’ 21.5

  10. 2026-10-04degraded: openrouter

    1 new model, open gap 4.4

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Opus 5.5 57.6 ยท gap 4.4

    • New: Ling 3.1 Flash (41.1)

  11. 2026-10-03

    1 score move, open gap 4.4

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Opus 5.5 57.6 ยท gap 4.4

    • NVIDIA Nemotron 3 Nano 4B Intelligence 7.4 โ†’ 4.8

  12. 2026-10-01

    2 new models, 3 price changes, open gap 4.4

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Opus 5.5 57.6 ยท gap 4.4

    • New: Gemini 4 Argon (52.6)
    • New: Solar Mini 4 (24.1)
    • Nemotron 3.5 Lightning Blended $/M $0.11 โ†’ $0.10
    • Nemotron 3.5 Lightning In $/M $0.07 โ†’ $0.06
    • Nemotron 3.5 Lightning Out $/M $0.22 โ†’ $0.20

  13. 2026-09-30

    1 new model, 5 score moves, open gap 4.4

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Opus 5.5 57.6 ยท gap 4.4

    • New: GPT-6.1 Sol (51.8)
    • JT-4.1 Flash 236B A21B Intelligence 27.3 โ†’ 33.9
    • Ling 3.0 Flash Intelligence 24.9 โ†’ 20.1
    • Gemma 4 31B Intelligence 19.0 โ†’ 14.7
    • Ling 3.0 Tiny Intelligence 15.3 โ†’ 11.1
    • Inkling Small Intelligence 27.8 โ†’ 25.7

  14. 2026-09-29

    1 new model, 1 score move, 6 price changes, open gap 4.4

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Opus 5.5 57.6 ยท gap 4.4

    • New: Claude Sonnet 5.5 (56.0)
    • Qwen3.5 9B Intelligence 13.7 โ†’ 11.2
    • Granite 4.1 8B Blended $/M $0.06 โ†’ $0.00
    • Granite 4.1 8B In $/M $0.05 โ†’ $0.00
    • Granite 4.1 8B Out $/M $0.10 โ†’ $0.00
    • Hy3 Blended $/M $0.25 โ†’ $0.24
    • Hy3 In $/M $0.14 โ†’ $0.14
    • Hy3 Out $/M $0.58 โ†’ $0.56

  15. 2026-09-28

    2 new models, open gap 4.4

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Opus 5.5 57.6 ยท gap 4.4

  16. 2026-09-25

    1 score move, 5 price changes, open gap 4.4

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Opus 5.5 57.6 ยท gap 4.4

    • Apodex 1.1 Intelligence 30.4 โ†’ 26.4
    • Gemma 4 26B A4B Blended $/M $0.16 โ†’ $0.17
    • Gemma 4 26B A4B Out $/M $0.34 โ†’ $0.37
    • Muse Glimmer Blended $/M $0.64 โ†’ $0.58
    • Muse Glimmer In $/M $0.35 โ†’ $0.33
    • Muse Glimmer Out $/M $1.50 โ†’ $1.35

  17. 2026-09-24

    2 new models, 5 price changes, 1 newly runnable, open gap 4.4

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Opus 5.5 57.6 ยท gap 4.4

    • Gemma 4 26B A4B Blended $/M $0.18 โ†’ $0.16
    • Gemma 4 26B A4B In $/M $0.12 โ†’ $0.10
    • Gemma 4 26B A4B Out $/M $0.37 โ†’ $0.34
    • Trinity Large Thinking Blended $/M $0.41 โ†’ $0.39
    • Trinity Large Thinking Out $/M $0.90 โ†’ $0.80
    • Ling-3.0-flash-VL now mirrored here

  18. 2026-09-23

    4 new models, 11 price changes, open gap 4.4

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Opus 5.5 57.6 ยท gap 4.4

    • New: Claude Opus 5.5 (57.6)
    • New: GPT-6 Sol (47.5)
    • New: MiMo-V2.6-Pro (46.3)
    • New: GPT-6 Luna (37.3)
    • GLM-5.1 Blended $/M $2.00 โ†’ $1.98
    • GLM-5.1 In $/M $1.20 โ†’ $1.28
    • GLM-5.1 Out $/M $4.40 โ†’ $4.07
    • Ling-3.0-flash-Fin Blended $/M $0.00 โ†’ $0.11
    • Ling-3.0-flash-Fin In $/M $0.00 โ†’ $0.07
    • Ling-3.0-flash-Fin Out $/M $0.00 โ†’ $0.22
    • Nemotron 3 Ultra 550B A55B Blended $/M $1.05 โ†’ $1.07
    • Nemotron 3 Ultra 550B A55B Out $/M $2.40 โ†’ $2.50
    • Nemotron 3.5 Lightning Blended $/M $0.10 โ†’ $0.11
    • Nemotron 3.5 Lightning In $/M $0.06 โ†’ $0.07
    • Nemotron 3.5 Lightning Out $/M $0.20 โ†’ $0.22

  19. 2026-09-22

    2 new models, 9 price changes, open gap 0.2

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Fable 5.1 53.4 ยท gap 0.2

    • New: Grok 4.7 (46.4)
    • New: MiMo-V2.6-Pro (46.3)
    • Ling-3.0-flash-VL Blended $/M $0.00 โ†’ $0.11
    • Ling-3.0-flash-VL In $/M $0.00 โ†’ $0.07
    • Ling-3.0-flash-VL Out $/M $0.00 โ†’ $0.22
    • Nemotron 3 Nano Omni 30B A3B Reasoning Blended $/M $0.16 โ†’ $0.42
    • Nemotron 3 Nano Omni 30B A3B Reasoning In $/M $0.09 โ†’ $0.20
    • Nemotron 3 Nano Omni 30B A3B Reasoning Out $/M $0.36 โ†’ $1.09
    • Nemotron 3 Super 120B A12B Blended $/M $0.31 โ†’ $0.45
    • Nemotron 3 Super 120B A12B In $/M $0.19 โ†’ $0.30
    • Nemotron 3 Super 120B A12B Out $/M $0.65 โ†’ $0.90

  20. 2026-09-20

    7 score moves, open gap 0.2

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Fable 5.1 53.4 ยท gap 0.2

    • Ling 3.0 Flash Intelligence 20.6 โ†’ 24.9
    • Gemma 4 31B Intelligence 15.4 โ†’ 19.0
    • Ling 3.0 Tiny Intelligence 11.9 โ†’ 15.3
    • MiMo-V2.5 Intelligence 22.3 โ†’ 25.2
    • Mercury 2 Intelligence 11.5 โ†’ 13.8
    • Inkling Small Intelligence 26.1 โ†’ 27.8
    • Agnes 2.5 Pro Alpha Intelligence 27.8 โ†’ 26.8

  21. 2026-09-19

    1 new model, 1 newly runnable, open gap 0.2

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Fable 5.1 53.4 ยท gap 0.2

    • New: Step 5 Preview (43.6)
    • GLM 5.3 Flash now mirrored here

  22. 2026-09-17

    1 new model, 3 price changes, open gap 0.2

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Fable 5.1 53.4 ยท gap 0.2

    • New: Ling-3.0-flash-Fin (23.0)
    • HyperNova 60B 2605 (high, based on gpt-oss-120b) Blended $/M $0.07 โ†’ $0.00
    • HyperNova 60B 2605 (high, based on gpt-oss-120b) In $/M $0.04 โ†’ $0.00
    • HyperNova 60B 2605 (high, based on gpt-oss-120b) Out $/M $0.14 โ†’ $0.00

  23. 2026-09-16

    1 new model, 3 score moves, 3 price changes, open gap 0.2

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Fable 5.1 53.4 ยท gap 0.2

    • New: Qwen3.8 Max (40.3)
    • Qwen3.8 Max (0902) Intelligence 40.3 โ†’ 45.4
    • Qwen3.8 Max (0902) Coding 71.8 โ†’ 76.2
    • North Mini Code Intelligence 12.8 โ†’ 9.9
    • Nex-N2-Pro Blended $/M $1.00 โ†’ $0.00
    • Nex-N2-Pro In $/M $0.50 โ†’ $0.00
    • Nex-N2-Pro Out $/M $2.50 โ†’ $0.00

  24. 2026-09-15

    2 score moves, 5 price changes, open gap 0.2

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Fable 5.1 53.4 ยท gap 0.2

    • Ling 3.0 Flash Intelligence 24.9 โ†’ 20.6
    • Kimi K2.6 Intelligence 31.3 โ†’ 27.5
    • Hy3 Blended $/M $0.24 โ†’ $0.25
    • Hy3 In $/M $0.14 โ†’ $0.14
    • Hy3 Out $/M $0.55 โ†’ $0.58
    • Nemotron 3 Ultra 550B A55B Blended $/M $1.10 โ†’ $1.05
    • Nemotron 3 Ultra 550B A55B Out $/M $2.60 โ†’ $2.40

  25. 2026-09-14

    4 new models, 1 score move, 3 price changes, open gap 0.2

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Fable 5.1 53.4 ยท gap 0.2

    • New: K2 Horizon MoVA 36B A4B (25.7)
    • New: K2 Horizon 7B (21.0)
    • New: K2 Horizon 3.7B (16.2)
    • New: K2 Horizon 0.9B (3.0)
    • Command A Intelligence 7.0 โ†’ 13.9
    • Command A Blended $/M $4.38 โ†’ $0.00
    • Command A In $/M $2.50 โ†’ $0.00
    • Command A Out $/M $10.0 โ†’ $0.00

  26. 2026-09-12

    1 new model, 3 score moves, 1 newly runnable, open gap 0.2

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Fable 5.1 53.4 ยท gap 0.2

    • New: Agnes 3.0 Flash (35.5)
    • GLM-5.2 Intelligence 38.6 โ†’ 34.0
    • K2 Horizon 375B A23B Intelligence 33.9 โ†’ 30.7
    • Agnes 2.5 Pro Alpha Intelligence 26.8 โ†’ 27.8
    • Ling-3.0-flash-VL now in Chat

  27. 2026-09-11

    2 new models, 4 score moves, 2 top-10 changes, 3 price changes, open gap 0.2

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Fable 5.1 53.4 ยท gap 0.2

    • Qwen3.6 35B A3B Intelligence 22.3 โ†’ 18.8
    • Qwen3.8-Flash-Next Intelligence 42.2 โ†’ 39.9
    • MiniCPM5-2B Intelligence 14.3 โ†’ 13.1
    • GLM-5.1 Intelligence 27.4 โ†’ 26.4
    • DeepSeek V4.1 Flash entered open top 10 on Intelligence (#5)
    • GLM 5 left open top 10 on Intelligence
    • LongCat 2.0 Blended $/M $1.30 โ†’ $0.53
    • LongCat 2.0 In $/M $0.75 โ†’ $0.30
    • LongCat 2.0 Out $/M $2.95 โ†’ $1.20

  28. 2026-09-09

    164 score moves, 6 price changes, open gap 0.2

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Fable 5.1 53.4 ยท gap 0.2

    • GPT-5.4 nano Intelligence 30.9 โ†’ 21.2
    • Qwen3.5 122B A10B Intelligence 24.8 โ†’ 16.2
    • Mistral Medium 3.5 Intelligence 22.7 โ†’ 14.9
    • Qwen3.8 27B Intelligence 41.4 โ†’ 33.9
    • GPT-5.4 mini Intelligence 31.9 โ†’ 24.6
    • Quasar 438B (max, based on GLM-5.2) Intelligence 34.2 โ†’ 27.1
    • Qwen3.6 27B Intelligence 29.0 โ†’ 21.9
    • GPT-5.5 Intelligence 45.6 โ†’ 38.6
    • Muse Spark 1.2 Intelligence 46.8 โ†’ 39.8
    • Qwen3.5 397B A17B Intelligence 26.1 โ†’ 19.1
    • Muse Spark 1.1 Intelligence 41.2 โ†’ 34.3
    • MiniMax-M2.7 Intelligence 30.1 โ†’ 23.2
    • Gemma 4 31B Intelligence 22.2 โ†’ 15.4
    • Claude Sonnet 5 Intelligence 45.1 โ†’ 38.4
    • Gemini 3.5 Flash Intelligence 39.7 โ†’ 33.0
    • Inkling Intelligence 32.2 โ†’ 25.5
    • Qwen3.7 Max Intelligence 36.6 โ†’ 29.9
    • Qwen3.8 2.4T A95B Intelligence 46.7 โ†’ 40.0
    • Qwen3.8 Max Intelligence 46.9 โ†’ 40.3
    • Hy3 Intelligence 32.4 โ†’ 25.8
    • Kimi K3 Intelligence 50.2 โ†’ 43.8
    • Kimi K2.7 Code Intelligence 32.7 โ†’ 26.3
    • Grok 4.5 Intelligence 45.5 โ†’ 39.1
    • Ring-2.6-1T Intelligence 23.7 โ†’ 17.3
    • Gemini 3.1 Pro Preview Intelligence 36.7 โ†’ 30.4
    • DeepSeek V4 Flash 0731 Intelligence 40.8 โ†’ 34.5
    • Muse Glimmer Intelligence 24.4 โ†’ 18.1
    • Grok 4.6 Intelligence 50.6 โ†’ 44.4
    • MiMo-V2.5-Pro Intelligence 32.6 โ†’ 26.4
    • Inkling Small Intelligence 32.2 โ†’ 26.1
    • LongCat 2.0 Intelligence 25.8 โ†’ 19.7
    • MiniMax-M3 Intelligence 35.7 โ†’ 29.6
    • DeepSeek V4 Flash Intelligence 30.7 โ†’ 24.6
    • Qwen3.7 Plus Intelligence 31.9 โ†’ 25.8
    • Gemini 3.6 Flash Intelligence 40.3 โ†’ 34.3
    • Nemotron 3 Ultra 550B A55B Intelligence 29.3 โ†’ 23.4
    • Gemini 3.8 Flash Intelligence 47.1 โ†’ 41.2
    • GPT-5.6 Luna Intelligence 43.4 โ†’ 37.5
    • MiMo-V2.5 Intelligence 28.2 โ†’ 22.3
    • DeepSeek V4 Pro 0813 Intelligence 42.1 โ†’ 36.3
    • Gemini 3.7 Flash Intelligence 45.2 โ†’ 39.4
    • Claude Opus 4.8 Intelligence 47.8 โ†’ 42.0
    • DeepSeek V4 Flash Vision Intelligence 40.7 โ†’ 35.0
    • Granite 4.1 8B Intelligence 1.1 โ†’ 6.6
    • Olmo 3.1 32B Instruct Intelligence 1.0 โ†’ 6.5
    • Sarvam 30B Intelligence 1.1 โ†’ 6.6
    • Phi-4 Mini Instruct Intelligence 1.0 โ†’ 6.3
    • GPT-5.5 Instant (June 2026) Intelligence 21.6 โ†’ 26.8
    • Qwen3.5 0.8B Intelligence 1.0 โ†’ 6.1
    • Nemotron 3 Super 120B A12B Intelligence 18.6 โ†’ 13.6
    • Gemini 3.5 Flash-Lite Intelligence 27.6 โ†’ 22.7
    • Command A+ Intelligence 2.1 โ†’ 7.0
    • Granite 4.1 3B Intelligence 1.0 โ†’ 5.9
    • LFM2 24B A2B Intelligence 1.0 โ†’ 5.9
    • Qwen3.5 2B Intelligence 2.1 โ†’ 6.9
    • Phi-4 Multimodal Instruct Intelligence 1.0 โ†’ 5.8
    • Muse Spark 1.3 Intelligence 53.0 โ†’ 48.2
    • MiniCPM-V 4.6 1.3B Intelligence 1.0 โ†’ 5.7
    • Nex-N2-Pro Intelligence 32.7 โ†’ 28.2
    • Solar Pro 4 Intelligence 32.7 โ†’ 28.2
    • Apodex 1.1 Intelligence 34.9 โ†’ 30.4
    • DeepSeek R1 Distill Qwen 1.5B Intelligence 1.0 โ†’ 5.5
    • GLM-5 Intelligence 32.4 โ†’ 27.9
    • GLM-5.1 Intelligence 31.9 โ†’ 27.4
    • GPT-5.2 Codex Intelligence 33.0 โ†’ 28.5
    • GPT-5.6 Terra Intelligence 46.8 โ†’ 42.3
    • Grok 4.3 Intelligence 29.9 โ†’ 25.4
    • Grok Build 0.1 0616 Intelligence 31.7 โ†’ 27.2
    • JT-4.1 Flash 236B A21B Intelligence 31.8 โ†’ 27.3
    • LFM2.5-8B-A1B Intelligence 2.7 โ†’ 7.2
    • MiMo-V2-Pro Intelligence 33.1 โ†’ 28.6
    • Motif 3 (Beta) Intelligence 36.8 โ†’ 32.3
    • Qwen3.6 Max Preview Intelligence 32.9 โ†’ 28.4
    • Qwen3.6 Plus Intelligence 31.5 โ†’ 27.0
    • Kimi K2.6 Intelligence 35.8 โ†’ 31.3
    • Muse Spark Intelligence 35.8 โ†’ 31.3
    • Claude Opus 4.6 Intelligence 30.8 โ†’ 26.4
    • Agnes 2.5 Pro Alpha Intelligence 31.2 โ†’ 26.8
    • GLM-5-Turbo Intelligence 31.0 โ†’ 26.6
    • GPT-5.3 Codex Intelligence 36.9 โ†’ 32.5
    • Motif 3 Intelligence 38.0 โ†’ 33.6
    • GLM 5.3 Flash Intelligence 46.2 โ†’ 41.9
    • Claude Sonnet 4.6 Intelligence 29.0 โ†’ 24.7
    • Grok 4.20 0309 Intelligence 29.5 โ†’ 25.2
    • Grok 4.20 0309 v2 Intelligence 30.0 โ†’ 25.7
    • NVIDIA Nemotron 3 Nano 4B Intelligence 3.1 โ†’ 7.4
    • Solar Open2 250B Intelligence 29.0 โ†’ 24.7
    • DeepSeek Coder V2 Lite Instruct Intelligence 1.0 โ†’ 5.3
    • Agnes 2.5 Pro Beta Intelligence 39.5 โ†’ 35.2
    • MiMo-V2-Omni-0327 Intelligence 29.4 โ†’ 25.1
    • MiMo-V2-Omni Intelligence 28.1 โ†’ 23.9
    • Granite 4.1 30B Intelligence 3.2 โ†’ 7.4
    • LFM2.5-1.2B-Instruct Intelligence 1.0 โ†’ 5.2
    • LFM2.5-1.2B-Thinking Intelligence 1.0 โ†’ 5.2
    • GPT-5.6 Sol Intelligence 51.3 โ†’ 47.1
    • A.X-K2 Intelligence 26.8 โ†’ 22.7
    • GLM 5V Turbo Intelligence 27.6 โ†’ 23.5
    • Hy3-preview Intelligence 26.8 โ†’ 22.7
    • Kimi K2.5 Intelligence 27.6 โ†’ 23.5
    • GPT-5.5 Instant (May 2026) Intelligence 26.7 โ†’ 22.7
    • MiniMax-M2.5 Intelligence 26.8 โ†’ 22.8
    • Molmo2-8B Intelligence 1.0 โ†’ 5.0
    • Qwen3 Coder Next Intelligence 14.1 โ†’ 10.1
    • Qwen3.5 27B Intelligence 26.9 โ†’ 22.9
    • KAT Coder Pro V2 Intelligence 25.6 โ†’ 21.7
    • Qwen3 1.7B Intelligence 1.0 โ†’ 4.9
    • Step3 VL 10B Intelligence 3.8 โ†’ 7.7
    • G9v3-39A5B Intelligence 25.7 โ†’ 21.8
    • K2 Horizon 375B A23B Intelligence 37.8 โ†’ 33.9
    • Qwen3.6 35B A3B Intelligence 26.2 โ†’ 22.3
    • Gemma 4 E2B Intelligence 4.0 โ†’ 7.8
    • LFM2.5-VL-1.6B Intelligence 1.0 โ†’ 4.8
    • Qwen3 0.6B Intelligence 1.0 โ†’ 4.8
    • Tiny Aya Global Intelligence 1.0 โ†’ 4.8
    • Ling 3.0 Tiny Intelligence 15.7 โ†’ 11.9
    • GPT-5.4 Intelligence 42.8 โ†’ 39.0
    • GLM-5.3 Intelligence 48.6 โ†’ 44.9
    • Falcon-H1R-7B Intelligence 4.1 โ†’ 7.8
    • Qwen3 Max Thinking Intelligence 25.0 โ†’ 21.3
    • DeepSeek V4 Pro Intelligence 34.5 โ†’ 30.9
    • Qwen3.5 Omni Plus Intelligence 24.0 โ†’ 20.4
    • Claude Opus 4.7 Intelligence 44.3 โ†’ 40.7
    • Claude Fable 5 Intelligence 53.2 โ†’ 49.7
    • GLM-5.2 Intelligence 42.1 โ†’ 38.6
    • K-EXAONE 2.0 Intelligence 23.1 โ†’ 19.7
    • Mercury 2 Intelligence 14.9 โ†’ 11.5
    • Claude Fable 5.1 Intelligence 56.8 โ†’ 53.4
    • Claude Opus 5 Intelligence 54.1 โ†’ 50.7
    • Qwen3.8-Flash-Next Intelligence 45.6 โ†’ 42.2
    • Step 3.7 Flash Intelligence 22.9 โ†’ 19.5
    • Gemini 3.1 Flash-Lite Intelligence 19.3 โ†’ 16.0
    • Qwen3.5 35B A3B Intelligence 22.6 โ†’ 19.3
    • DeepSeek V4 Flash Intelligence 22.1 โ†’ 18.9
    • JT-35B-Flash Intelligence 21.8 โ†’ 18.7
    • DeepSeek R1 Distill Qwen 32B Intelligence 5.3 โ†’ 8.4
    • LFM2.5-2.6B Intelligence 5.3 โ†’ 8.4
    • Nanbeige4.1-3B Intelligence 5.3 โ†’ 8.4
    • Nemotron 3.5 Lightning Intelligence 16.4 โ†’ 13.6
    • Ling-2.6-1T Intelligence 19.6 โ†’ 17.0
    • MiniCPM5-1B Intelligence 6.2 โ†’ 8.8
    • Qwen3 4B 2507 Intelligence 6.2 โ†’ 8.8
    • Sarvam 105B Intelligence 6.2 โ†’ 8.8
    • Gemma 4 26B A4B Intelligence 19.2 โ†’ 16.7
    • Gemma 4 E4B Intelligence 6.4 โ†’ 8.9
    • Ling 3.0 Flash Intelligence 27.4 โ†’ 24.9
    • Step 3.5 Flash Intelligence 19.1 โ†’ 16.6
    • Step 3.5 Flash 2603 Intelligence 19.5 โ†’ 17.0
    • Tri-21B-Think Intelligence 6.6 โ†’ 9.0
    • Trinity Large Thinking Intelligence 13.1 โ†’ 10.9
    • QwQ 32B Intelligence 7.5 โ†’ 9.5
    • GPT-6 Astra Intelligence 54.7 โ†’ 52.8
    • DiffusionGemma 26B A4B Intelligence 7.6 โ†’ 9.5
    • Tri-21B-think Preview Intelligence 7.7 โ†’ 9.6
    • Ling 2.6 Flash Intelligence 7.9 โ†’ 9.7
    • GLM-4.7-Flash Intelligence 16.6 โ†’ 14.9
    • Granite 4.2 30B Intelligence 16.5 โ†’ 14.8
    • Solar Pro 3 Intelligence 9.4 โ†’ 7.8
    • Granite 4.2 3B Intelligence 10.7 โ†’ 9.1
    • Granite 4.2 8B Intelligence 13.9 โ†’ 12.4
    • Gemma 4 12B Intelligence 15.6 โ†’ 14.2
    • Mistral Small 4 Intelligence 12.8 โ†’ 11.5
    • Nemotron 3 Nano Omni 30B A3B Reasoning Intelligence 9.0 โ†’ 10.3
    • Celeris-1 Intelligence 7.6 โ†’ 6.3
    • Qwen3.5 9B Intelligence 14.8 โ†’ 13.7
    • Nemotron 3 Nano Omni 30B A3B Reasoning Blended $/M $0.13 โ†’ $0.16
    • Nemotron 3 Nano Omni 30B A3B Reasoning In $/M $0.07 โ†’ $0.09
    • Nemotron 3 Nano Omni 30B A3B Reasoning Out $/M $0.30 โ†’ $0.36
    • Nemotron 3 Super 120B A12B Blended $/M $0.35 โ†’ $0.31
    • Nemotron 3 Super 120B A12B In $/M $0.20 โ†’ $0.19
    • Nemotron 3 Super 120B A12B Out $/M $0.80 โ†’ $0.65

  29. 2026-09-08

    1 new model, 9 score moves, 2 top-10 changes, open gap 3.6

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Fable 5.1 56.8 ยท gap 3.6

    • New: MiniCPM5-2B (15.0)
    • DeepSeek V4 Flash Intelligence 33.1 โ†’ 30.7
    • Muse Spark 1.1 Intelligence 43.3 โ†’ 41.2
    • Claude Opus 4.8 Intelligence 46.4 โ†’ 47.8
    • MiMo-V2.5 Intelligence 29.5 โ†’ 28.2
    • Celeris-1 Intelligence 6.3 โ†’ 7.6
    • DeepSeek V4 Pro Intelligence 35.8 โ†’ 34.5
    • Kimi K2.7 Code Intelligence 33.9 โ†’ 32.7
    • Trinity Large Thinking Intelligence 12.0 โ†’ 13.1
    • Qwen3.7 Plus Intelligence 30.8 โ†’ 31.9
    • GLM-5 entered open top 10 on Intelligence (#10)
    • Hy3 left open top 10 on Intelligence

  30. 2026-09-06

    1 new model, 2 score moves, 4 top-10 changes, 1 newly runnable, open gap 3.6

    Intelligence: DeepSeek V4 Pro 0423 53.2 vs frontier Claude Fable 5.1 56.8 ยท gap 3.6

    • Left the board: GLM-5.3-Flash
    • Gemini 3.5 Flash Intelligence 41.9 โ†’ 39.7
    • Gemini 3.1 Pro Preview Intelligence 38.3 โ†’ 36.7
    • GLM 5.3 Flash entered open top 10 on Intelligence (#4)
    • GLM-5 left open top 10 on Intelligence
    • GLM 5.3 Flash entered open top 10 on Coding (#2)
    • Inkling Small left open top 10 on Coding
    • DeepSeek V4 Flash 0731 now mirrored here