Epoch AI: open-weight models 6-7 months behind on cyber; GLM-5.3 tops Cyber ECI at 160
Epoch AI finds the open-weight cyber frontier trails the closed frontier by 6-7 months: Z.ai's GLM-5.3 scores 160 on its Cyber ECI, beating GPT-5.4 but trailing Claude Mythos Preview and GPT-5.5. Dean Ball responds that during AI research acceleration a constant 6-7 month lag represents a growing subjective gap.
All posts · 3, most important first
- How far behind the closed frontier are open-weight cyber capabilities? About 6–7 months. Zai’s GLM-5.3 is the highest-scoring open-weight model on our Cyber ECI. Its score of 160 beats OpenAI’s GPT-5.4, released in March.
35 likes · 8 reposts - Just a reminder that being consistently behind in stable objective time (“6-9 months”) during a time of substantial AI-enabled research acceleration is in fact being behind by a growing amount. A 6-month lag today is subjectively larger than a 6-month lag 6 months ago.@epochairesearch: How far behind the closed frontier are open-weight cyber capabilities? About 6–7 months. Zai’s GLM-5.3 is the highest-scoring open-weight model on our Cyber ECI. Its score of 160 beats OpenAI’s GPT-5.4, released in March. open on X39 likes · 2 reposts
- Read the full Data Insight and methodology: epoch.ai/data-insights/…4 likes · 0 reposts
@epochairesearch org · 
@deanwball policy ·