GLM-5.3 Released: Coding Nears Claude Fable 5, Security Focus, Delivers Full Game in Test
量子位 · wechat · 2026-08-14
Zhipu releases GLM-5.3, topping open-source benchmarks with coding and agent abilities close to Claude Fable 5, and excelling in security (84.5% on CyberGym white-box code review, making it the strongest open-source security model). Pre-release red-teaming with Tsinghua, Nankai, etc. found 2,404 vulnerabilities. In tests, GLM-5.3 autonomously completed the Lord of the Rings benchmark (28 min), a 3D watch deconstruction demo (10+ min), and developed a deliverable simulation game (40 min) with frontend, backend, database, and permissions. In security tests, it found and fixed 12 vulnerability classes, passing 54/54 regression tests; without a GPU, it distinguished VERIFIED, INFERRED, and UNVERIFIED.
More from coding & agent
- Dev vibe-codes open-source newsletter platform, 83% cheaper than Beehiiv from $19/mo — marclou · 2026-10-15
- ERP AI launches Proto: 'headless SaaS' business software built for AI agents — Aiden_Tech_Ai · 2026-10-15
- FLUX.3 Image Klein ships open in Diffusers: layout control, 2s per image on H100 — ariG23498 · 2026-10-15
- Engineering managers need deep reading, not just agent summaries: 5 recommended resources — rseroter · 2026-10-15
- Adding motion graphics with just a prompt: the Claude Motion x ElevenLabs MCP workflow — CurieuxExplorer · 2026-10-15
- A $15k AI SEO client-finding system built on Claude, free to download — aryanXmahajan · 2026-10-15