Long Context Reasoning on AA-LCR
70.2ScoreKimi-K2.6
Evaluation Results
| Method | Links | |
|---|---|---|
| Kimi-K2.6Parameter Count=1T-A32B2026.06 | 70.2 | |
| MiniMax-2.7Parameter Count=230B-A10B2026.06 | 69.8 | |
| Qwen-3.5Parameter Count=397B-17B2026.06 | 68.3 | |
| DS-v4-ProParameter Count=1.6T-A49B2026.06 | 67.3 | |
| GLM-5.1Parameter Count=744B-A40B2026.06 | 66.9 | |
| Nemotron 3 UltraParameter Count=550B-A55B2026.06 | 65.4 | |
| DS-v4-FlashParameter Count=284B-A13B2026.06 | 62.7 | |
| LoongRLBackbone=Qwen3-30B-A3B-Thinking-25072026.05 | 53.5 | |
| LONGTRACERLBackbone=Qwen3-30B-A3B-Thinking-25072026.05 | 53.5 | |
| DocQABackbone=Qwen3-30B-A3B-Thinking-25072026.05 | 50.2 | |
| LongRLVRBackbone=Qwen3-30B-A3B-Thinking-25072026.05 | 48.5 | |
| gpt-oss-120bNumber of Parameters=117B, Optimization Level=high2026.01 | 48.3 | |
| LONGTRACERL-GRPOBackbone=Qwen3-30B-A3B-Thinking-25072026.05 | 48.2 | |
| BaseBackbone=Qwen3-30B-A3B-Thinking-25072026.05 | 47 | |
| K-EXAONEMode=Non-Reasoning, Architecture=MoE, Total Params=236B, Activated Params=23B2026.01 | 45.2 | |
| gpt-oss-120bNumber of Parameters=117B, Optimization Level=medium2026.01 | 45 | |
| LONGTRACERLBackbone=Qwen3-4B-Thinking-25072026.05 | 41.8 | |
| LongRLVRBackbone=Qwen3-4B-Thinking-25072026.05 | 37.5 | |
| GLM-4.5-AirNumber of Parameters=110B2026.01 | 37.3 | |
| Solar OpenNumber of Parameters=102B2026.01 | 35 | |
| LONGTRACERL-GRPOBackbone=Qwen3-4B-Thinking-25072026.05 | 34 | |
| BaseBackbone=Qwen3-4B-Thinking-25072026.05 | 33.2 | |
| DeepSeek-V3.2Mode=Non-Reasoning, Architecture=MoE, Total Params=671B, Activated Params=37B2026.01 | 32 | |
| LoongRLBackbone=Qwen3-4B-Thinking-25072026.05 | 32 | |
| Qwen3-235B-A22B Instruct-2507Architecture=MoE, Total Params=235B, Activated Params=22B2026.01 | 31.2 | |
| DocQABackbone=Qwen3-4B-Thinking-25072026.05 | 28.8 | |
| Qwen3.5-2B2026.06 | 25.6 | |
| Cosmos3-Edge2026.06 | 22.8 | |
| LONGTRACERL-GRPOBackbone=DeepSeek-R1-0528-Qwen3-8B2026.05 | 15 | |
| LONGTRACERLBackbone=DeepSeek-R1-0528-Qwen3-8B2026.05 | 15 | |
| BaseBackbone=DeepSeek-R1-0528-Qwen3-8B2026.05 | 13.8 | |
| LongRLVRBackbone=DeepSeek-R1-0528-Qwen3-8B2026.05 | 12.2 | |
| LoongRLBackbone=DeepSeek-R1-0528-Qwen3-8B2026.05 | 10.2 | |
| DocQABackbone=DeepSeek-R1-0528-Qwen3-8B2026.05 | 9.5 | |
| EXAONE 4.0Mode=Non-Reasoning, Architecture=Dense, Total Params=32B, Activated Params=32B2026.01 | 8 |