CodeGeeX Models by Zhipu AI
Last refreshed 2026-10-08. Next refresh: weekly.
Details
About
CodeGeeX is a multilingual code generation model from Tsinghua University's KEG group and Zhipu AI. The original model has 13 billion parameters and was pre-trained on code in more than 20 programming languages; by June 22, 2022 it had been trained on more than 850 billion tokens on 1,536 Huawei Ascend 910 processors. It can write code in languages such as Python, C++, Java, JavaScript, and Go, and translate code between them. The project also introduced the HumanEval-X benchmark for testing code generation across languages.
Decision facts
- Best fit
- coding
- Capability starting point
- CodeGeeX 13B with 4K context
- Lowest tracked input
- Not tracked
- Closest related family
- AutoGLM
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
Use when the workload needs 4K context and 13B parameters.
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| CodeGeeX 13B | Use when the workload needs 4K context and 13B parameters. | 2023-05 | 4K context13B parameters | Current |
Release Timeline
1 release groupSpecifications(1 models)
| Model | Released | Context | Parameters |
|---|---|---|---|
| CodeGeeX 13B | 2023-05 | 4K | 13B |



