Artificial Analysis

Critical Physics Tasks

An Artificial Analysis metric for research-level physics reasoning.

In index
Reasoning
Source ↗

Models215

Top result35.4%Claude Fable 5.2

Top-3 spread3.1 pts

Year2026

Result and price

143 of 215 results

Results

215 results

#ModelResult

01Claude Fable 5.235.4%

02Muse Wonder FullAbove fit32.9%

03GPT-5.6 SolAbove fit32.3%

04Claude Opus 5.5Above fit31.7%

04GPT-6 AstraAbove fit31.7%

04GPT-6.1 SolAbove fit31.7%

07Claude Sonnet 5.5Above fit31.4%

08GPT-6 SolAbove fit30.9%

09GPT-5.5 ProAbove fit30.6%

10GPT-5.4 ProAbove fit30.0%

10GPT-5.6 TerraAbove fit30.0%

12Claude Fable 5.1Above fit29.7%

13Claude Opus 5Above fit29.1%

14Claude Fable 5Above fit28.6%

15Muse WonderAbove fit28.3%

16GPT-6.1 Terra28.0%

17Gemini 4 Argon27.1%

17GPT-5.5Above fit27.1%

19MiMo-V2.6-ProAbove fit26.6%

20Gemini 3 Pro Deep Think25.7%

21Muse Spark 1.324.9%

22Kimi K3.1 Pro24.6%

23Kimi K323.4%

23GPT-5.4Above fit23.4%

25MiniMax M3.1 Flash22.3%

26Step 5 Preview20.9%

26Claude Opus 4.820.9%

26GLM-5.2Above fit20.9%

29GPT-5.6 Luna20.6%

30Sage-A1.520.0%

31GPT-6 Luna19.4%

32GLM-5.319.1%

33Claude Haiku 5.518.9%

34Gemini 3.8 Flash18.3%

35Ling 3.1 Flash18.0%

35DeepSeek V4 Pro 081318.0%

37Grok 4.717.7%

37Muse Spark 1.217.7%

37Qwen3.8 Max Preview17.7%

37Gemini 3.1 Pro17.7%

41Beam 1.0 501B-A23B FP817.4%

42Grok 4.617.1%

43Hy4 preview16.9%

43Claude Sonnet 516.9%

43GPT-5.3 Codex16.9%

46DeepSeek V4 Flash 073116.6%

47Beam 1.0 501B-A23B16.3%

48Grok 4.515.4%

48Beam 1.0 501B-A23B NVFP415.4%

50Muse Spark 1.115.1%

51Muse Spark Open14.9%

52DeepSeek V4.1 Flash 101714.3%

52Gemini 3.7 Flash14.3%

54Qwen3.7 Max13.4%

55Gemini 3.5 Flash13.1%

56Claude Opus 4.6 (Adaptive)12.6%

57MiMo-V2.6-Flash12.0%

57Claude Opus 4.7 (Adaptive)12.0%

59GPT-5.211.6%

60Muse Spark11.3%

61Qwen3.8-Flash-Next11.1%

62Nemotron 3.5 Super VL 120B A12B10.9%

63Gemini 3.6 Flash10.6%

63Nemotron 3.5 Super VL 120B A12B NVFP410.6%

65Mistral Large 410.0%

65Kimi K2.7 Code10.0%

65GPT-5.4 mini10.0%

68Quasar 438B9.4%

69GPT-5.4 nano9.3%

70Gemini 3 Pro9.1%

70Qwen3.7 Plus9.1%

72A.X K28.9%

73GPT-5.2-Codex8.7%

74Inkling-Small8.3%

75Kimi K2.68.0%

77Solar Pro 46.3%

78GPT-5.1-Codex-Max5.7%

78GPT-5 (high)5.7%

78GPT-5.1-Codex5.7%

81Qwen3.8-27B5.4%

83Claude Opus 4.75.1%

84Hy3 Preview4.9%

84Hy34.9%

87Apodex 1.14.6%

87Claude Opus 4.5 Thinking4.6%

87Apodex 1.1 Mini4.6%

91Nemotron 3.5 Bolt 74B-A9B4.3%

92MiMo-V2.5-Pro4.0%

93Qwen 3.6 Max (preview)3.7%

93MiniMax M33.7%

95Kimi K2.5 (Reasoning)3.1%

95Kimi K2.53.1%

95Nemotron 3 Ultra3.1%

98Qwen3.6 Plus2.9%

98Grok 4.1 Fast (Reasoning)2.9%

98Grok 4 Fast (Reasoning)2.9%

101Claude Opus 4.62.8%

102Muse Glimmer 30B2.6%

102Gemini 2.5 Pro2.6%

104Step 3.7 Flash2.3%

105GLM-52.0%

105Grok 42.0%

105DeepSeek V3.1 (Reasoning)2.0%

105Ling 3.0 Flash VL2.0%

109Qwen3.5 397B (Reasoning)1.7%

109Qwen3.5 397B1.7%

109Ling 3.0 Flash1.7%

109GLM-4.71.7%

113Gemini 3 Flash1.4%

113Gemma 4 31B1.4%

113GPT-OSS 20B1.4%

113DeepSeek-R11.4%

113Gemini 2.5 Flash1.4%

118Qwen3.6-27B1.1%

118o31.1%

118MiMo-V2-Omni1.1%

118GPT-OSS 120B1.1%

118K-Exaone1.1%

118Claude 4 Sonnet1.1%

124Claude Sonnet 4.60.9%

124Qwen3.5-27B0.9%

124Qwen3.5-35B-A3B0.9%

124K-EXAONE 2.00.9%

124DeepSeek V3.20.9%

124Trinity-Large-Thinking0.9%

124Trinity-Large-Preview0.9%

124Claude 3.7 Sonnet Thinking0.9%

132Qwen3.5-122B-A10B0.6%

132MiniMax M2.70.6%

132GLM-5V-Turbo0.6%

132o4-mini0.6%

132Gemma 4 E4B0.6%

137Claude Opus 4.50.3%

137GLM-5-Turbo0.3%

137MiMo-V2-Pro0.3%

137Qwen3.6-35B-A3B0.3%

137o10.3%

137Granite 4.2 30B0.3%

137Command A+0.3%

137North Mini Code0.3%

137Mistral Small 40.3%

137Granite 4.2 8B0.3%

137MiniCPM5-2B0.3%

137Magistral Medium0.3%

137Sarvam 30B0.3%

—Mistral Small 4 (Reasoning)0.3%

150GPT-5 (medium)0.0%

150Gemini 3.5 Flash-Lite0.0%

150Mistral Medium 3.5 128B0.0%

150Claude 4.1 Opus Thinking0.0%

150Mercury 2.50.0%

150Qwen3 Max0.0%

150Gemma 4 26B A4B0.0%

150Qwen3.5-4B0.0%

150MiMo-V2-Flash0.0%

150GLM-4.60.0%

150Grok Code Fast 10.0%

150Kimi K20.0%

150Solar Pro 30.0%

150Gemma 4 12B0.0%

150Nemotron 3 Nano Omni 30B A3B0.0%

150GPT-4.10.0%

150DeepSeek V3.10.0%

150Claude 3.7 Sonnet0.0%

150DeepSeek V3 03240.0%

150Nemotron 3.5 Lightning 30B A3B NVFP40.0%

150Ling 2.6 Flash0.0%

150Sarvam 105B0.0%

150Nemotron Ultra 253B0.0%

150GPT-4.1 mini0.0%

150Llama 4 Maverick0.0%

150Mistral Large 30.0%

150Celeris-10.0%

150Grok 4.1 Fast0.0%

150Granite 4.2 3B0.0%

150Nemotron 3 Nano 30B0.0%

150Exaone 4.0 32B0.0%

150Magistral Small0.0%

150LFM2.5-8B-A1B0.0%

150DeepSeek V30.0%

150Gemini 2.0 Flash0.0%

150LFM2.5-2.6B0.0%

150GPT-4o0.0%

150Llama 3.3 70B0.0%

150Devstral Small0.0%

150Mistral Large 20.0%

150Mistral Medium 30.0%

150Gemini 2.5 Flash-Lite0.0%

150Phi-4-reasoning0.0%

150Llama 3.1 405B0.0%

150Llama 4 Scout0.0%

150GPT-4.1 nano0.0%

150Nova Pro0.0%

150Gemma 3 27B0.0%

150Llama 3.1 70B0.0%

150Mistral Small 3.10.0%

150Claude 3.5 Haiku0.0%

150Phi-40.0%

150Claude 3 Haiku0.0%

150Gemma 4 E2B0.0%

150Granite-4.0-H-1B0.0%

150Granite-4.0-H-350M0.0%

150Granite-4.0-350M0.0%

150Exaone 4.0 1.2B0.0%

150GLM-4.5-Air0.0%

150LFM2.5-VL-1.6B-Extract0.0%

150Ling 3.0 Tiny0.0%

150Qwen3-Omni-30B-A3B-Instruct0.0%

150Qwen3-Omni-30B-A3B-Thinking0.0%

150Solar Pro 20.0%

150Ultravox v0.6 Llama 3.3 70B0.0%