←── back to feed
/topics/glm-5-2-frontier-model-release
GLM-5.2 frontier model release
19 items●3 sources●updated 5h ago●trend 5
Z.ai released GLM-5.2, an MIT-licensed open-source model designed for long-horizon software engineering tasks, achieving near-parity with proprietary frontier models on coding benchmarks. The model supports a 1M-token context window with up to 131K output tokens and ranks just behind Claude Opus 4.8 on FrontierSWE while edging out GPT-5.5.
- Scores 1% behind Claude Opus 4.8 and 1% ahead of GPT-5.5 on FrontierSWE long-horizon coding benchmark
- Supports 1M-token context window with up to 131,072 output tokens for agentic workflows
- MIT-licensed open weights model, positioning it as cost-effective alternative to proprietary coding models
- Marked substantial capability leap over predecessor GLM-5.1 in long-horizon task performance
- Available via unified Model API on Z.ai platform and as local deployment option
[HN]hacker news10
How GLM-5.2 beat Fable 5 At Website Design
MiniMax M3 vs. GLM 5.2: Codegen comparison across autonomous coding tasks
GPT-5.5 hallucinates 3x more than MIT-licensed GLM-5.2
GLM-5.2 – How to Run Locally
Running glm 5.2 on a Mac studio
GLM-5.2: The Most Powerful Open Model yet and the Brutal Reality of Running It
GLM 5.2 is now available via a unified Model API
GLM-5.2 is probably the most powerful text-only open weights LLM
GLM-5.2: Frontier Intelligence, Open Weights
GLM-5.2: Built for Long-Horizon Tasks
[BLG]blog/rss6
[AINews] GLM > GPT? GLM-5.2 passes vibe check; Z.ai forecasts Open Fable by December
zai-org/GLM-5
Hierarchical Attention via Domain Decomposition
Z.ai pitches GLM-5.2 for long-running software engineering tasks
GLM-5.2: Built for Long-Horizon Tasks
[AINews] GLM-5.2: the top Frontend Coding model in the world, IndexShare for Speculative Decoding
[BSKY]bluesky3
Credit to GLM-5.2 Max, the new Chinese open weights model, for pulling this off.
It's hard to pinpoint open-closed gap and so-on, but I trust the Arena team and just look where GLM 5.2 is on this. An MIT licensed, to be open weight model. At this point you could argue they have a better agent than Gemini does. That's a…
New podcast with my (former) colleague Finbarr Timbers on the state of frontier post-training recipes and how we got here. A fun review of how we went from InstructGPT (RLHF) -> DeepSeek R1 (reasoning) -> Today (agentic models w/ on-policy…