Z.ai · Model release
GLM 5.3: Coding Plan flagship, API and weights still pending
GLM 5.3 is the current GLM Coding Plan default — same base as 5.2, with documented post-training gains. Direct token pricing and open weights are not published yet.
What changed
- Z.ai released GLM 5.3 on 2026-08-14 for every Coding Plan subscriber; 5.2/5.1 requests on that plan route to 5.3.
- Official docs: 1M context, 128K max output, thinking always on, reasoning_effort low / high / max (default max).
- Direct `glm-5.3` API is coming soon and is absent from the per-token table; open weights are promised about two weeks after launch.
Best fit
- Existing GLM Coding Plan teams
- Long-horizon coding in supported harnesses
- Evaluating Z.ai's current flagship
What to be careful about
- Do not copy GLM 5.2's $1.40/$4.40 token rates onto 5.3.
- Vendor DeepSWE 66.9 used mini-swe-agent at 400K context — it is not the public Datacurve v1.1 board.
- Thinking cannot be disabled; a `disabled` request fails.
Dataset snapshot: GLM 5.3 is also represented in the checked comparison dataset. Prices and benchmark figures carry their own field dates.