DeepSeek V4 Pro scores 46.9 on the Respan Index (#29), ahead of GPT-6 Luna at 46.8 (#30). DeepSeek V4 Pro leads on 14 of the 20 benchmarks both report. GPT-6 Luna is 9.9x cheaper per token at list price. GPT-6 Luna has the larger context window (1.1M tokens).
* Estimated: no published scores in that category. How the Respan Index works
| GPQA Diamond |
| 90.9% |
| 90.5% |
| LiveBench Agentic Coding (LiveBench) | 54.9% | 51.2% |
| LiveBench Coding (LiveBench) | 77.2% | 79% |
| LiveBench Data Analysis (LiveBench) | 79.2% | 73.4% |
| LiveBench Instruction Following (LiveBench) | 67.7% | 55.9% |
| LiveBench Language (LiveBench) | 82.1% | 73.8% |
| LiveBench Mathematics (LiveBench) | 95.1% | 89.1% |
| LiveBench Reasoning (LiveBench) | 85.8% | 81.8% |
| Mystery Game Puzzles (Epoch AI) | 43% | 7% |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 96.7% | 98.9% |
| SimpleQA Verified | 47% | 41.4% |
| Terminal-Bench 4.0 (Vals AI) | 14.1% | 13.6% |
| LMArena Agent (LMArena) | 0.0119 | 0.0135 |
| LMArena Elo | 1451 | 1443 |
| LMArena WebDev (LMArena) | 1583 | 1579 |
| DeepSWE v1.1 | 62.7% | 66.6% |
Higher is better. Each score links to where it was published.