Compressed Query Delegation Evaluation on Benchmark B (N=200)
4.9G_EGPT-5.2
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GPT-5.2Latency (ms)=9502026.01 | 4.9 | 0.89 | |
| Claude 3.7 OpusLatency (ms)=11002026.01 | 4.7 | 0.91 | |
| Gemini 2.0 UltraLatency (ms)=7802026.01 | 4.6 | 0.88 | |
| Claude 3.5 SonnetLatency (ms)=8902026.01 | 4.2 | 0.84 | |
| GPT-4oLatency (ms)=5502026.01 | 4 | 0.76 | |
| Mistral Large 2Latency (ms)=6202026.01 | 3.8 | 0.72 | |
| Llama 3 70BLatency (ms)=4502026.01 | 3.1 | 0.68 | |
| Human baseline2026.01 | 1.2 | 0.12 |