Qwen3.8 27B scores 40.5 on the Respan Index (#45), ahead of GLM-5.2 at 40.1 (#48). GLM-5.2 leads on 9 of the 15 benchmarks both report. Qwen3.8 27B is 1.9x cheaper per token at list price. GLM-5.2 has the larger context window (1M tokens).
* Estimated: no published scores in that category. How the Respan Index works
| 73.7% |
| 76.6% |
| LiveBench Instruction Following (LiveBench) | 62.3% | 72.7% |
| LiveBench Language (LiveBench) | 76.2% | 74.3% |
| LiveBench Mathematics (LiveBench) | 89.8% | 86.2% |
| LiveBench Reasoning (LiveBench) | 78.6% | 80% |
| LMArena Agent (LMArena) | 0.0348 | -0.0166 |
| LMArena Elo | 1476 | 1439 |
| LMArena WebDev (LMArena) | 1605 | 1591 |
| Terminal-Bench 2.1 | 81% | 73% |
| SWE-Bench Pro | 62.1% | 61.7% |
| SimpleBench (SimpleBench) | 58.8% | 60.2% |
Higher is better. Each score links to where it was published.