🏆 HF Hub Benchmark Dashboard
Last updated: 2026-07-10T10:02:28 UTC · auto-refreshes every 6h
38 benchmarks
345 models
686 entries
34 active
4 empty
🧠 Knowledge · 3 benchmarks · 184 models
MMLU-Pro
50 entries↗ View on Hub
| Model | Score | In $/1M | Out $/1M | Context | TTFT | Throughput | License | Params | Providers | |
|---|---|---|---|---|---|---|---|---|---|---|
| 🥇 | MiniMaxAI/MiniMax-M2.1 | 88 | $0.30 | $1.20 | 204.8K | 1,288 ms | 53 t/s | other | 229B | novitafeatherless-ai |
| 🥈 | internlm/Intern-S2-Preview | 88 | — | — | — | — | — | apache-2.0 | 36B | — |
| 🥉 | Qwen/Qwen3.5-397B-A17B | 87.8 | $0.45 | $3.00 | 262.1K | 266 ms | 163 t/s | apache-2.0 | 403B | novitafeatherless-aiscaleway +2 |
| 4 | deepseek-ai/DeepSeek-V4-Pro | 87.5 | $1.30 | $2.60 | 1.0M | 475 ms | 60 t/s | mit | 862B | novitatogetherfireworks-ai +1 |
| 5 | moonshotai/Kimi-K2.5 | 87.1 | $0.45 | $2.25 | 262.1K | 838 ms | 31 t/s | other | 1059B | novitafeatherless-aideepinfra |
| 6 | nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16 | 86.8 | $1.00 | $5.00 | 262.1K | 811 ms | 119 t/s | other | 561B | deepinfra |
| 7 | nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4 | 86.8 | $0.60 | $2.40 | 512.3K | 260 ms | 136 t/s | other | 335B | togetherfireworks-ai |
| 8 | Qwen/Qwen3.5-122B-A10B | 86.7 | $0.29 | $2.40 | 262.1K | 590 ms | 98 t/s | apache-2.0 | 125B | novitadeepinfra |
| 9 | deepseek-ai/DeepSeek-V4-Flash | 86.4 | $0.09 | $0.18 | 1.0M | 571 ms | 107 t/s | mit | 158B | novitafireworks-aifeatherless-ai +1 |
| 10 | Qwen/Qwen3.6-27B | 86.2 | $0.32 | $3.19 | 262.1K | 392 ms | 70 t/s | apache-2.0 | 28B | featherless-aiovhclouddeepinfra |
| 11 | Qwen/Qwen3.5-27B | 86.1 | $0.26 | $2.40 | 262.1K | 271 ms | 78 t/s | apache-2.0 | 28B | novitafeatherless-aideepinfra |
| 12 | zai-org/GLM-5 | 86 | $0.60 | $2.08 | 202.8K | 1,387 ms | 38 t/s | mit | 754B | novitafeatherless-aizai-org +1 |
| 13 | Qwen/Qwen3.6-35B-A3B | 85.2 | $0.15 | $0.95 | 262.1K | 347 ms | 105 t/s | apache-2.0 | 36B | featherless-aiscalewaydeepinfra |
| 14 | deepseek-ai/DeepSeek-R1-0528 | 85 | $0.50 | $2.15 | 163.8K | 576 ms | 25 t/s | mit | 685B | novitafeatherless-aideepinfra |
| 15 | zai-org/GLM-4.5 | 84.6 | — | — | — | 2,288 ms | 54 t/s | mit | 358B | zai-org |
| 16 | stepfun-ai/Step-3.5-Flash | 84.4 | $0.10 | $0.30 | 262.1K | 424 ms | 155 t/s | apache-2.0 | 199B | featherless-aideepinfra |
| 17 | deepseek-ai/DeepSeek-R1 | 84 | $0.70 | $2.50 | 64K | 894 ms | 28 t/s | mit | 685B | novita |
| 18 | LGAI-EXAONE/K-EXAONE-236B-A23B | 83.8 | — | — | — | — | — | other | 237B | — |
| 19 | nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16 | 83.73 | — | — | — | — | — | other | 124B | — |
| 20 | RedHatAI/NVIDIA-Nemotron-3-Super-120B-A12B-BF16 | 83.73 | — | — | — | — | — | other | 124B | — |
| 21 | exolabs/NVIDIA-Nemotron-3-Super-120B-A12B-UD-Q3_K_M-dequant-bf16-vllm | 83.73 | — | — | — | — | — | other | 121B | — |
| 22 | exolabs/NVIDIA-Nemotron-3-Super-120B-A12B-UD-Q4_K_M-dequant-bf16-vllm | 83.73 | — | — | — | — | — | other | 121B | — |
| 23 | exolabs/NVIDIA-Nemotron-3-Super-120B-A12B-MXFP4_MOE-dequant-bf16-vllm | 83.73 | — | — | — | — | — | other | 121B | — |
| 24 | internlm/Intern-S1 | 83.5 | — | — | — | — | — | apache-2.0 | 241B | — |
| 25 | LGAI-EXAONE/EXAONE-4.5-33B | 83.3 | — | — | — | — | — | other | 34B | — |
| 26 | Qwen/Qwen3-235B-A22B-Instruct-2507 | 83 | $0.09 | $0.10 | 262.1K | 225 ms | 79 t/s | apache-2.0 | 235B | novitanscaletogether +2 |
| 27 | ByteDance-Seed/Seed-OSS-36B-Instruct | 82.7 | — | — | — | — | — | apache-2.0 | 36B | — |
| 28 | meituan-longcat/LongCat-Flash-Chat | 82.7 | — | — | — | — | — | mit | 562B | — |
| 29 | MiniMaxAI/MiniMax-M2 | 82 | $0.30 | $1.20 | 204.8K | 1,795 ms | 46 t/s | other | 229B | novitafeatherless-ai |
| 30 | skt/A.X-K1 | 81.5 | — | — | — | — | — | apache-2.0 | 519B | — |
| 31 | zai-org/GLM-4.5-Air | 81.4 | $0.13 | $0.85 | 131.1K | 908 ms | 49 t/s | mit | 110B | novitazai-org |
| 32 | deepseek-ai/DeepSeek-V3-0324 | 81.2 | $0.27 | $1.12 | 163.8K | 1,389 ms | 33 t/s | mit | 685B | novitafeatherless-ai |
| 33 | MiniMaxAI/MiniMax-M1-40k | 81.1 | — | — | — | — | — | apache-2.0 | 456B | — |
| 34 | jdopensource/JoyAI-LLM-Flash | 81.02 | — | — | — | — | — | — | 49B | — |
| 35 | moonshotai/Kimi-K2-Instruct | 81 | $0.57 | $2.30 | 131.1K | 1,006 ms | 39 t/s | other | 1026B | novitafeatherless-ai |
| 36 | Qwen/Qwen3-30B-A3B-Thinking-2507 | 80.9 | — | — | — | — | — | apache-2.0 | 31B | — |
| 37 | openai/gpt-oss-120b | 80.8 | $0.04 | $0.17 | 131.1K | 214 ms | 784 t/s | apache-2.0 | 120B | groqnovitacerebras +7 |
| 38 | MiniMaxAI/MiniMax-M2.5 | 80.1 | $0.15 | $1.15 | 204.8K | 608 ms | 78 t/s | other | 229B | novitafeatherless-aideepinfra |
| 39 | baidu/ERNIE-4.5-300B-A47B-PT | 78.4 | — | — | — | — | — | apache-2.0 | 300B | — |
| 40 | meituan-longcat/LongCat-Flash-Lite | 78.29 | — | — | — | — | — | mit | 69B | — |
| 41 | MiniMaxAI/MiniMax-Text-01 | 75.7 | — | — | — | — | — | — | 456B | — |
| 42 | openai/gpt-oss-20b | 73.6 | $0.03 | $0.14 | 131.1K | 218 ms | 702 t/s | apache-2.0 | 22B | groqnovitanscale +5 |
| 43 | Qwen/Qwen2.5-72B | 71.59 | — | — | — | — | — | other | 73B | — |
| 44 | microsoft/phi-4 | 70.4 | $0.07 | $0.14 | 16.4K | 267 ms | 72 t/s | mit | 15B | featherless-aideepinfra |
| 45 | Qwen/Qwen3-4B-Instruct-2507 | 69.6 | $0.01 | $0.03 | 262.1K | 586 ms | 125 t/s | apache-2.0 | 4B | nscalefeatherless-ai |
| 46 | baidu/ERNIE-4.5-300B-A47B-Base-PT | 69.5 | — | — | — | — | — | apache-2.0 | 299B | — |
| 47 | Qwen/Qwen2.5-32B | 69.23 | — | — | — | — | — | apache-2.0 | 33B | — |
| 48 | XiaomiMiMo/MiMo-V2.5-Pro | 68.5 | $1.00 | $3.00 | 1.0M | 369 ms | 52 t/s | mit | 1023B | deepinfra |
| 49 | Qwen/Qwen3-235B-A22B | 68.18 | $0.20 | $0.60 | 41.0K | 801 ms | 19 t/s | apache-2.0 | 235B | novitanscalefeatherless-ai |
| 50 | mistralai/Mistral-Large-Instruct-2411 | 67.94 | — | — | — | — | — | other | 123B | — |