GLM 5.2 is now available on Vercel's AI Gateway, accessible via the AI SDK with model identifier zai/glm-5.2. The headline upgrade: context window expanded to 1M tokens, a 5x increase over GLM 5.1's 200K limit.
The model targets long-horizon engineering tasks, maintaining project-level context across a single task run and applying engineering standards more consistently than its predecessor. That reliability claim, and what it means for agentic workloads, is what makes the full changelog worth reading.
AI Gateway wraps the model with a unified API, built-in Zero Data Retention, per-key budgets, failover, and BYOK support, all at provider pricing with no platform fee. GLM 5.2 joins a broader model leaderboard and a live playground for direct comparison.
[READ ORIGINAL →]