Multi-discipline Knowledge Evaluation on CodaSet ID MMLU-PRO (test)
84.86AccuracyQwen3-235B-A22B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Qwen3-235B-A22B2026.05 | 84.86 | 5.8 | |
| DeepSeek-V3.12026.05 | 83.94 | 5.2 | |
| DeepSeek-V3.2-Exp2026.05 | 83.76 | 4 | |
| MODEL-SAT2026.05 | 81.82 | 6 | |
| Kimi-K2-Instruct2026.05 | 81.75 | 11.6 | |
| DecoRfallback_model=DeepSeek-V3.12026.05 | 80.36 | 2.2 | |
| gpt-oss-120b2026.05 | 80.21 | 2.8 | |
| RouterDC2026.05 | 79.99 | 4.4 | |
| DecoRfallback_model=gpt-oss-120b2026.05 | 79.77 | 2 | |
| EmbedLLM2026.05 | 79.74 | 4.3 | |
| Random Router2026.05 | 77.92 | 4.8 | |
| KNN Router2026.05 | 77.51 | 2.5 | |
| Mistral-Small-3.2-24B2026.05 | 73.77 | 1.9 | |
| gemma-3-27b-it2026.05 | 71.58 | 3.5 | |
| gemma-3-12b-it2026.05 | 69.64 | 1 | |
| LLM Router2026.05 | 65.56 | 1.1 |