· Updated
GLM-5.1 as a flagship open-source model for agentic coding (AI that autonomously writes and iterates on code). It ranks first in open source and third globally on SWE-Bench Pro, a benchmark (standardized test for measuring model capabilities), scoring 58.4.The update shifts the focus from simple code generation to long-horizon autonomy. While most models struggle with multi-step reasoning over time, this model is built to run for eight hours straight. It uses a self-review loop to refine features and interactions, effectively bridging the gap between open-weights models and top-tier proprietary systems.
You can use the model for complex technical workflows like database optimization, where it has demonstrated a 6x performance boost over standard 50-turn sessions. GLM-5.1 is available now through OpenRouter, Vercel AI Gateway, and Requesty, with model weights hosted on Hugging Face for local deployment.
Introducing GLM-5.1: The Next Level of Open Source - Top-Tier Performance: #1 in open source and #3 globally across SWE-Bench Pro, Terminal-Bench, and NL2Repo. - Built for Long-Horizon Tasks: Runs autonomously for 8 hours, refining strategies through thousands of iterations. https://t.co/YQZLhKVwik
More like this
Zhipu AIApr 1
OpenCodeApr 8
Fireworks AIApr 28
KimiApr 24