Penyedia AI

43 laboratorium dan penyedia di lima papan peringkat independen.

Claude Mythos Preview97.4

65 model⌀ 48.2

16 model⌀ 52.13 bobot terbuka

GPT-5.6 Sol80.8

109 model⌀ 41.62 bobot terbuka

Qwen3.8 Max79.8

90 model⌀ 36.553 bobot terbuka

Gemini 3.1 Pro Preview74.0

65 model⌀ 42.314 bobot terbuka

Grok 4.20 Beta174.0

31 model⌀ 50.0

Deepseek V4 Pro High Preview73.0

33 model⌀ 45.419 bobot terbuka

Glm 5.2 Max73.0

23 model⌀ 51.711 bobot terbuka

Ernie 5.0 Preview 120372.0

5 model⌀ 54.6

Longcat Flash Chat 2602 Exp72.0

2 model⌀ 71.0

Amazon Nova Experimental Chat 26 02 1071.0

25 model⌀ 35.5

Nvidia Nemotron 3 Ultra 550b A55b Nvfp471.0

17 model⌀ 26.31 bobot terbuka

Hunyuan Hy3 Preview71.0

13 model⌀ 59.4

Minimax M2.570.0

9 model⌀ 41.24 bobot terbuka

Mistral Medium 250870.0

34 model⌀ 29.415 bobot terbuka

Step 3.5 Flash70.0

6 model⌀ 54.5

Muse Spark 1.169.9

43 model⌀ 38.518 bobot terbuka

Command A 03 202568.0

12 model⌀ 39.9

Olmo 3.1 32b Instruct67.0

7 model⌀ 54.51 bobot terbuka

Ling Flash 2.067.0

5 model⌀ 34.0

Granite 4.1 8b65.0

7 model⌀ 43.92 bobot terbuka

2 model⌀ 43.5

Phi 3 Medium 4k Instruct60.0

10 model⌀ 46.62 bobot terbuka

Sarvam 105B50.0

3 model⌀ 49.73 bobot terbuka

MiMo V2.5 Pro48.0

4 model⌀ 36.64 bobot terbuka

1 model⌀ 44.4

Thinking Machines

Inkling Small41.7

2 model⌀ 36.7

4 model⌀ 17.7

Seed 2.0 Pro40.0

2 model⌀ 29.3

Agnes 2 5 Pro Alpha39.7

1 model⌀ 39.7

Ring 2 6 1t31.3

1 model⌀ 31.3

K2 Think V222.0

1 model⌀ 22.0

Apriel V1 5 15b Thinker21.6

2 model⌀ 21.2

North Mini Code20.2

1 model⌀ 20.2

Trinity Large Thinking18.4

1 model⌀ 18.4

1 model⌀ 12.4

Jamba 1 5 Large4.8

Apertus 70b Instruct2.0

Async Flash V1 50.0

Cogito V2 1 Reasoning0.0

1 model⌀ 0.01 bobot terbuka