GLM-Legal Release - PLawBench: leaderboard
Publisher-reported PLawBench rubric score in the October 13 GLM-Legal release, covering Chinese legal consultation, case analysis and document drafting. Five native chart models; higher is better. Exact subset, version, judge, aggregation, effort and evaluation harness are unreported. Keep this source-qualified comparison separate from benchmarklist and Vals runs.
Metric: Reported rubric score. Source: autoclaw.z.ai. Saturation forecast: Around February 2028. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Fable 5 | 77.7 | #20 |
| 2 | Kimi K3 | 76.2 | #35 |
| 3 | GLM-5.3 Flash | 73.8 | #77 |
| 4 | GPT-6 Astra | 69.1 | #6 |
No result here: #3 Claude Opus 5.5, #7 GPT-6.1 Sol, #9 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=glm-legal-release-plawbench · How It Works · Data refreshed daily, snapshot 2026-11-05.