GLM-5.3 Ties Kimi K3 as Leading Open Weights Model, Ranking 2nd Globally in Agentic Work
Models44d agoZ AI has launched the GLM-5.3 API, a model designed for coding, defensive cybersecurity, and agentic tasks that ties Kimi K3 as the most intelligent open weights model with a score of 60 on the Artificial Analysis Intelligence Index. The model ranks second among all evaluated AI for real world agentic knowledge work, posting a GDPval-AA v2 Elo of 1770, behind only Claude Opus 5. GLM-5.3 also places third overall on Design Arena with an Elo of 1351, the second highest rank for any open weights model in design tasks.
The model retains the 753B total parameter and 40B active parameter size of GLM-5.2, with performance gains derived from post-training in realistic engineering environments. Z AI plans to release the model weights within a week. While the API is priced at $1.40 per 1M input tokens and $4.40 per 1M output tokens, GLM-5.3 uses roughly 20% more output tokens per task than its predecessor.
Key sources
- SOURCE@zai_org“Built for coding, defensive cybersecurity, and long-horizon agentic tasks”x.com
- SOURCE@zixuanli_“beginning a broader review of the model weights as we work toward a responsible open-weight release”x.com
- SOURCE@zixuanli_“deliver significant improvements in reasoning, general chat, and specialized domains such as law and finance”x.com
- SUPPORT@designarena“2nd-highest-ranked open-weight model on real-world design tasks”x.com
- SUPPORT@artificialanlys“19% cheaper per task compared to Kimi K3 ($0.84) and 45% cheaper than GPT-5.6 Sol ($1.23)”x.com