GLM 5.1: what developers actually say
We read every Hacker News comment from Nov 2025 – Nov 2026 that gave a real opinion on GLM 5.1. 54 of the 67 opinions come from people who say they use it themselves.
Developers give GLM 5.1 a 78% approval score across 67 opinions from 50 authors, and 54 of those opinions come from direct users. Code quality is the strongest theme at 28 positive mentions against 4 negative, and cost follows at 18 positive with none negative. Monthly approval fell from 95 in May to 68 in June on 19 opinions, so the recent direction is down. July holds only 1 opinion, too few to call a recovery.
Pick it if
- Want strong code quality at a low price
- Are running long agent tasks on a tight budget
- Want a model that works well in OpenCode or a spec-driven workflow
Look elsewhere if
- Need stable behavior past 200k tokens of context
- Depend on security work, which it often refuses
- Need fast responses during peak hours
Strengths and weaknesses
Each row counts opinions that mentioned the topic. Bars to the right of the line are praise, bars to the left are complaints.
How opinion has moved
What developers say
Praise
Complaints
Summaries are paraphrased. Each links to the original comment.
FAQ
Is GLM 5.1 worth it?
Most developers say yes. It scores 78% approval, and cost is the clearest strength with 18 positive mentions and none negative. The weak spot is reliability, where negative mentions outnumber positive ones 3 to 1.
Is GLM 5.1 better than Claude Opus?
Opinions split. Two people preferred it to Claude Opus 4.6, and some say it comes close on agent tasks at about 1/20 the price. Others describe it as clearly weaker and slower than Opus, so the answer depends on how much you weigh cost against raw capability.
How does GLM 5.1 handle long context?
One user reported performance degrading after about 200k tokens on the coding plan. Another described it as quirky near full context. Keep tasks inside a smaller window if reliability matters.