qwen3-32b-thinking
54±0
54interval55
aiqbench · LLM benchmarks
Benchmarks
- Sources
- 8
- Entries tracked
- 22,318
- Last sync
- 17 min17 minutes ago
All sources syncing normally
- Arena · Text· 17 minutes ago
- Arena · WebDev· 17 minutes ago
- Arena · Vision· 17 minutes ago
- Arena · Agent· 17 minutes ago
- GPQA Diamond· 55 minutes ago
- SWE-bench· 55 minutes ago
- Terminal-Bench· 55 minutes ago
- ARC-AGI-2· 55 minutes ago
Filter by org:
Released:
One grid, every column. Pick axes in the Axes panel — one axis ranks like a single board, several blend into a weighted composite, all of them read as the index. Rank is assigned before any filter and never renumbers.
| Model | Org | Interval | Swing | i | i | i | i | i | i | i | i | i | i | i | i | i | i | i | i | i | i | i | i | i | i | Add to compare | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 296 | qwen3-32b-thinking | Alibaba | — | — | 54.3 67.4Ranked 54.3 = (1 × 67.4 + 3 × 50) ÷ 4. 67.4 is the mean percentile over the 1 of 22 selected axes this model is measured on; fewer axes means a stronger pull toward 50. | — | — | 1/22low | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 67 | — | — | — | — | — | — | — |
Blend is coverage-shrunk; the raw composite is shown greyed beside it · A low badge means fewer than 3 boards scored this model