Zhipu GLM 5.3 Launches with Doubled Long-Horizon Agent Performance

ollama · x · 2026-08-29

Zhipu released GLM 5.3 and Flash versions. Built on GLM 5.2 architecture, GLM 5.3 significantly improves long-horizon agentic tasks via post-training, with SWE-Marathon scores doubling from 19.4 to 42.5. Coding capabilities surged, with Terminal-Bench 3.0 rising from 4.6 to 28.3. Hosted on Ollama cloud with 1M context.

Related event: Z.ai Open-Sources GLM-5.3, Touting Agentic Coding and Cyber Defense(13 posts)→

Original post →

More from coding & agent

coding & agent channel →