Zhipu's GLM-5.3 tops open-weight Cyber ECI, ~6-7 months behind frontier
dl_weekly · x · 2026-10-29
This week's Data Languages issue analyzes z.ai's GLM-5.3 on the Cyber ECI (cybersecurity eval): it tops the open-weight category at a score the closed-weight frontier reached roughly 6-7 months earlier — suggesting open-weight models now trail the frontier by only about half a year on cybersecurity agent capabilities.
More from Models
- Blogger: Opus 5.5 is frontier at writing quality with better research taste — tszzl · 2026-10-30
- OpenAI scraps Pro 200 limit cut and doubles usage limits across all Pro tiers — minchoi · 2026-10-30
- Altman: OpenAI Almost Ditched Token Pricing for an AWS-Style Bill, Users Refused — menhguin · 2026-10-30
- User math: $1 of Son costs 7x Opus on $20 plan, but tasks only 1.3x — alvelda · 2026-10-30
- Claude gets the right answer but can't prove time complexity — geoffreyirving · 2026-10-30
- Healthcare company's ops team burns 1.2B Opus tokens in a single day — RishabJainK · 2026-10-30