Z.ai (Zhipu) · MIT
GLM 5.3 API
Agentic engineering — long-horizon coding, terminal tasks, hybrid reasoning.
Context: up to 1M tokens · Released: August 2026
# GLM 5.3 via kryyt (OpenAI SDK) from openai import OpenAI client = OpenAI(base_url="https://api.kryyt.com/v1", api_key="$KRYYT_API_KEY") r = client.chat.completions.create( model="glm-5.3", messages=[{"role":"user","content":"Explain MoE routing in one paragraph."}], ) print(r.choices[0].message.content)
Overview
The GLM-5 series from Z.ai is a 744B-parameter mixture-of-experts family tuned for agentic engineering. GLM-5.3 leads on coding and cyber-defense workloads; GLM-5.3-Flash is an MIT-licensed fast tier; GLM-5.2 adds native multimodality and a 1M-token context.
Best for:
- Repo-scale coding & terminal agents
- Hybrid fast/deliberate reasoning
- Chinese + English tasks
GLM 5.3 benchmark performance
GLM-5.3's gains came entirely from post-training on the GLM-5.2 base, with the largest jump on the harder Terminal-Bench 3.0. Scores below mix GLM variants; each row names the model measured.
| Benchmark | Model | Score (%) |
|---|---|---|
| Terminal-Bench 2.1Agentic coding in real terminal environments | GLM-5.3 | |
| CyberGymSecurity vulnerability discovery | GLM-5.3 | |
| Terminal-Bench 2.1Agentic coding in real terminal environments | GLM-5.3 Flash | |
| DeepSWE v1.1Repairing fresh, real-world software issues | GLM-5.3 Flash | |
| SWE-bench ProResolving issues in professional codebases | GLM-5.2 | |
| Terminal-Bench 3.0Harder next-generation terminal tasks | GLM-5.3 |
Sources: GLM-5.3 benchmarks (MindStudio, citing Z.ai) (opens in new tab); GLM-5.3 benchmarks (Emergent, citing Z.ai) (opens in new tab); GLM-5.3-Flash benchmarks (DataCamp, citing Z.ai) (opens in new tab); Z.ai GLM-5 series repository (opens in new tab).
GLM-5.3 vs GLM-5.2
| Benchmark | Scores (%) |
|---|---|
| Terminal-Bench 3.0Harder next-generation terminal tasks |
Sources: GLM-5.3 benchmarks (MindStudio, citing Z.ai) (opens in new tab).
GLM on Terminal-Bench 2.1 across releases
View chart data as a table
| Model | Released | Terminal-Bench 2.1 (%) |
|---|---|---|
| GLM-5.1 | — | 62.0 |
| GLM-5.2 | Jun 2026 | 81.0 |
| GLM-5.3 | Aug 2026 | 88.2 |
Scores are vendor-reported unless marked independent, as of September 2026. Labs use different agent harnesses, reasoning settings, and benchmark versions, so treat cross-lab gaps of a few points as roughly equal and test on your own workload. See all models on the benchmarks page.
GLM 5.3 models & pricing
| Model | API ID | Context | Pricing | License | Status |
|---|---|---|---|---|---|
| GLM-5.3 | glm-5.3 | 200K | See account pricing | MIT | Soon |
| GLM-5.3 Flash | glm-5.3-flash | 1M | See account pricing | MIT | Soon |
| GLM-5.2 | glm-5.2 | 1M | See account pricing | MIT | Soon |
glm-5.3
- Model
- GLM-5.3
- Context
- 200K
- Pricing
- See account pricing
- License
- MIT
- Status
- Soon
glm-5.3-flash
- Model
- GLM-5.3 Flash
- Context
- 1M
- Pricing
- See account pricing
- License
- MIT
- Status
- Soon
glm-5.2
- Model
- GLM-5.2
- Context
- 1M
- Pricing
- See account pricing
- License
- MIT
- Status
- Soon
Usage is charged at Kryyt's published rates for the selected model and features. Review current pricing and usage in your account.
GLM 5.3 FAQ
What is GLM 5.3?
The GLM-5 series from Z.ai is a 744B-parameter mixture-of-experts family tuned for agentic engineering. GLM-5.3 leads on coding and cyber-defense workloads; GLM-5.3-Flash is an MIT-licensed fast tier; GLM-5.2 adds native multimodality and a 1M-token context.
How do I call GLM 5.3 on kryyt?
Send a POST request to https://api.kryyt.com/v1/chat/completions with your kryyt key and set "model" to one of: glm-5.3, glm-5.3-flash, glm-5.2.
What license is GLM 5.3 under?
GLM 5.3 is available under the MIT license via its provider, Z.ai (Zhipu).
What context window does GLM 5.3 support?
GLM 5.3 supports up to 1M tokens depending on the specific model variant.
Ship with one endpoint
Get a key, change your base URL, and call every model on this page. No new SDKs.