Long-context Reasoning on FRAMES
84.7ScoreClaude-Opus-4.5
Evaluation Results
| Method | Links | |
|---|---|---|
| Claude-Opus-4.52026.06 | 84.7 | |
| Seed2.0 Pro2026.06 | 84.5 | |
| GPT-5.2 High2026.06 | 84 | |
| Gemini-3-FlashTier=High2026.06 | 83.7 | |
| DocQABackbone=Qwen3-30B-A3B-Thinking-25072026.05 | 83.5 | |
| Seed2.0 Lite2026.06 | 83.4 | |
| GPT-5-miniTier=High2026.06 | 82.9 | |
| LoongRLBackbone=Qwen3-30B-A3B-Thinking-25072026.05 | 81.9 | |
| LONGTRACERLBackbone=Qwen3-30B-A3B-Thinking-25072026.05 | 81.9 | |
| Gemini-3-pro High2026.06 | 81.9 | |
| LongRLVRBackbone=Qwen3-30B-A3B-Thinking-25072026.05 | 81.7 | |
| BaseBackbone=Qwen3-30B-A3B-Thinking-25072026.05 | 80.7 | |
| Seed2.0 Mini2026.06 | 80.5 | |
| LONGTRACERL-GRPOBackbone=Qwen3-30B-A3B-Thinking-25072026.05 | 79.6 | |
| LONGTRACERLBackbone=Qwen3-4B-Thinking-25072026.05 | 79.5 | |
| Claude-Sonnet-4.52026.06 | 78.7 | |
| LongRLVRBackbone=Qwen3-4B-Thinking-25072026.05 | 78.5 | |
| DocQABackbone=Qwen3-4B-Thinking-25072026.05 | 78.3 | |
| BaseBackbone=Qwen3-4B-Thinking-25072026.05 | 76.7 | |
| LONGTRACERL-GRPOBackbone=Qwen3-4B-Thinking-25072026.05 | 76.1 | |
| LoongRLBackbone=Qwen3-4B-Thinking-25072026.05 | 75.8 | |
| LONGTRACERLBackbone=DeepSeek-R1-0528-Qwen3-8B2026.05 | 74.3 | |
| DocQABackbone=DeepSeek-R1-0528-Qwen3-8B2026.05 | 73.4 | |
| BaseBackbone=DeepSeek-R1-0528-Qwen3-8B2026.05 | 73.2 | |
| LONGTRACERL-GRPOBackbone=DeepSeek-R1-0528-Qwen3-8B2026.05 | 73.1 | |
| LoongRLBackbone=DeepSeek-R1-0528-Qwen3-8B2026.05 | 72.6 | |
| LongRLVRBackbone=DeepSeek-R1-0528-Qwen3-8B2026.05 | 70.3 |