Best open-weights AI models

133 models · Index. Published weights only. Method

Ranked models

November 2026

#ModelIndexOutput / 1MSpeed

01Kimi K3.1 Pro75.7$10.0043/s

02MiMo-V2.6-Pro70.2$0.8731/s

03MiniMax M3.1 Flash69.3$0.60100/s

04DeepSeek V4.1 Flash 101768.5$0.40261/s

05Step 5 Preview68.4$2.70105/s

06Ling 3.1 Flash67.2$0.13581/s

07Kimi K366.9$16.0044/s

08Hy4 preview66.4$2.5067/s

09GLM-5.365.8$1.4086/s

10Qwen3.8 Max65.6—41/s

11DeepSeek V4 Pro 081365.5$4.4442/s

12Muse Spark Open65.1$4.2586/s

13Qwen3.8-Flash-Next64.3—55/s

13MiMo-V2.6-Flash64.3$0.2828/s

15dots3-note Preview63.4Free—

16GLM-5.3-Flash63.1$0.078122/s

17DeepSeek V4 Flash 073161.9$0.3061/s

18Mistral Large 460.7$2.10113/s

19GLM-5.260.6$6.80111/s

20Sage-A1.560.5——

21DeepSeek V4 Flash Vision Exp60.4$1.3274/s

22Beam 1.0 501B-A23B60.2—345/s

23Beam 1.0 501B-A23B FP859.2——

23Qwen3.8-27B59.2$1.88113/s

25Beam 1.0 501B-A23B NVFP458.5——

26Kimi K2.658.2$3.4181/s

27Kimi K2.7 Code57.7$3.5035/s

28DeepSeek V4 Pro 042357.6$1.7889/s

29MiMo-V2.5-Pro56.3$1.8047/s

30GLM-5.156.0$4.40102/s

31MiniMax M355.7$1.2079/s

31Apodex 1.1 Mini55.7Free—

33Qwen3.5 397B (Reasoning)55.3$3.60—

33Step 3.7 Flash55.3$1.15112/s

35Inkling-Small55.1$1.2080/s

36MiMo-V2.554.8$2.004/s

36Inkling54.8$4.0585/s

38Qwen3.5 397B53.8$3.6059/s

39Solar Open 253.3——

40Muse Glimmer 30B53.0$1.2090/s

40Kimi K2.5 (Reasoning)53.0$3.00—

42Ripple-65B Preview52.3$0.6063/s

42Hy3 Preview52.3$0.6069/s

44Kimi K2.552.2$3.3236/s

44Ternary Bonsai 2 27B52.2$0.5021/s

46Qwen3.6-27B52.1$3.2060/s

47A.X K251.9——

47Qwen3.5-122B-A10B51.9$2.0870/s

49GLM-551.6$3.2045/s

50Qwen3.5-27B51.3$2.6069/s

51MiniMax M2.751.2$0.8435/s

52Nemotron 3.5 Super VL 120B A12B51.0——

53Ling 3.0 Flash50.9$0.06373/s

54Nemotron 3.5 Super VL 120B A12B NVFP450.8——

54Nemotron 3 Ultra50.8Free—

56Qwen3.6-35B-A3B50.6$0.9065/s

57Mistral Medium 3.5 128B50.3$7.50—

58MiniMax M2.550.2$1.0844/s

59GLM-4.750.0$2.2028/s

60Qwen3.5-35B-A3B49.4$1.0043/s

60DeepSeek V3.2 (Thinking)49.4$2.19—

62MiniMax M2.149.1$1.2034/s

63K-EXAONE 2.048.8——

64Spark-X2.6-4B48.7——

65Kimi K2 Thinking48.2$2.5048/s

66Gemma 4 31B47.9Free35/s

67Qwen3 235B 2507 (Reasoning)47.4$3.50108/s

68Agents-A1-4B46.8——

69Nemotron 3.5 Bolt 74B-A9B46.4——

70GLM-4.545.7$2.2025/s

71DeepSeek-R1-052844.6$2.1520/s

72Gemma 4 26B A4B44.5Free—

73Qwen3.5-4B44.3—20/s

74DeepSeek V3.244.2$0.3819/s

75Trinity-Large-Thinking43.5$0.80138/s

76MiMo-V2-Flash43.4—129/s

76GPT-OSS 120B43.4$0.6034/s

78DeepSeek V3.1 (Reasoning)43.0——

79Mistral Medium 3.542.7$7.5092/s

80Qwen3 235B 250742.3$0.8048/s

81Trinity-Large-Preview42.2$1.00—

82GLM-4.641.8$2.0030/s

83GPT-OSS 20B41.7$0.13310/s

84Granite 4.2 30B41.6——

85LFM3-1.7B41.5$0.20218/s

86Command A+40.2$1.50161/s

87LFM2.5-24B-A2B39.8$0.70280/s

88Kimi K239.6$2.3037/s

89LongCat-Flash-Lite-Sparse39.3——

90Gemma 4 12B39.0——

91DeepSeek-R138.9$2.5018/s

92Nemotron 3 Nano Omni 30B A3B38.6—228/s

93ZAYA1-8B38.2——

94Qwen3 235B-A22B38.1$1.8244/s

95North Mini Code38.0Free70/s

96DeepSeek V3.137.9—98/s

96Mistral Small 437.9$0.60114/s

98Granite 4.2 8B37.6$0.15101/s

99DeepSeek V3 032437.1——

99MiniCPM5-2B37.1——

101Nemotron 3.5 Lightning 30B A3B NVFP437.0——

101Ling 2.6 Flash37.0—104/s

103Sarvam 105B36.7—191/s

104QwQ-32B36.3——

105Nemotron Ultra 253B35.9——

106Llama 4 Maverick34.4$1.0061/s

107Mistral Large 334.3$1.5061/s

108Granite 4.2 3B33.2——

109Nemotron 3 Nano 30B32.9—270/s

110Exaone 4.0 32B32.4——

111Magistral Small31.4——

112LFM2.5-8B-A1B30.6—335/s

113Sarvam 30B30.3——

114DeepSeek V329.9$1.1037/s

115LFM2.5-2.6B29.3$0.50189/s

116Llama 3.3 70B28.2——

116Devstral Small28.2——

116Gemma 4 E4B28.2——

119Mistral Large 227.4$6.0070/s

120Phi-4-reasoning27.1—41/s

121Llama 3.1 405B27.0—29/s

122Llama 4 Scout26.8$0.3063/s

123Gemma 3 27B25.3—31/s

124Llama 3.1 70B24.8——

125LLaDA2.2-mini24.3——

126Mistral Small 3.124.2—28/s

127Phi-422.4$0.1493/s

128Gemma 4 E2B21.9——

129Llama 3.1 8B17.9——

130Granite-4.0-H-1B16.3——

131MiniCPM5-1B15.5——

132Granite-4.0-H-350M12.7——

133Granite-4.0-350M11.9——