Complex retrieval and positional sorting on MRCR 512K~1M
46.88ScoreGemini-2.5-Flash-Thinking
Evaluation results
| Method | Links | |
|---|---|---|
| Gemini-2.5-Flash-ThinkingEvaluation Framework=Full-context Inference2025.12 | 46.88 | |
| Gemini-2.5-ProEvaluation Framework=Full-context Inference2025.12 | 39.51 | |
| QwenLong-L1.5-30B-A3BEvaluation Framework=Memory Agent Framework2025.12 | 22.53 | |
| Qwen-Flash-Thinking-1MEvaluation Framework=Full-context Inference2025.12 | 9.97 | |
| Qwen3-30B-A3B-Thinking-2507Evaluation Framework=Memory Agent Framework2025.12 | 4.24 | |
| MemAgent-14BEvaluation Framework=Memory Agent Framework2025.12 | 3.11 |