Recommended Models

MaaS provides multiple models for you to use. You can easily integrate model services into your business by following tutorials or API documentation.

Recommended models

GLM-5.2	GLM-5.1	GLM-5
GLM-5.2 is Zhipu AI's next-generation flagship model, officially launched and open-sourced on June 17, 2026. Its core features include: Enhanced long-horizon task capability: reduces context drift and goal forgetting in complex tasks. State-of-the-art coding and long-horizon task evaluation: achieves open-source SOTA performance, delivering greater stability in complex system engineering and deep debugging. Improved real-world development experience: more reliable project-level context handling, adherence to engineering standards, and multi-platform development. Support for 198K sequence length, with maximum input of 192K, maximum output of 128K, and maximum thought chain length of 64K. Support for advanced functions, including thought chains, function calling, and prefix continuation.	GLM-5.1, the latest flagship model from Zhipu, delivers enhanced coding capabilities and superior performance on long-horizon tasks. It can autonomously operate for up to eight hours on a single task, covering the full cycle from planning and execution to iterative optimization, and producing engineering-level results. While GLM-5.1 matches Claude Opus 4.6 in general intelligence and raw coding proficiency, it significantly outperforms the global frontier in long-horizon sustained execution. It excels at autonomous, multi-stage tasks over extended periods, making it the ideal foundation for building highly resilient autonomous agents and long-horizon coding engines. It supports sequence lengths up to 198K, with a maximum input of 192K, a maximum output of 128K, and a maximum thought chain of 96K. It supports chain-of-thought, function calling, and prefix continuation.	Compared with GLM-4.5, GLM-5 expands total parameters from 355B to 744B (boosting active parameters from 32B to 40B) and increases pre-training data from 23T to 28.5T tokens. In addition, GLM-5 integrates the DeepSeek sparse attention (DSA) mechanism, which significantly reduces deployment costs while maintaining the long context capability. Compared to GLM-4.7, GLM-5 delivers substantial performance gains across a wide spectrum of academic benchmarks. It establishes a new state-of-the-art (SOTA) among global open-source models in reasoning, coding, and agentic tasks, further closing the capability gap with frontier closed-source models. It supports sequence lengths up to 198K, with a maximum input of 192K, a maximum output of 64K, and a maximum thought chain of 64K. It supports chain-of-thought, function calling, and prefix continuation.

GLM-5.2

GLM-5.1

GLM-5

GLM-5.2 is Zhipu AI's next-generation flagship model, officially launched and open-sourced on June 17, 2026. Its core features include:

Enhanced long-horizon task capability: reduces context drift and goal forgetting in complex tasks.

State-of-the-art coding and long-horizon task evaluation: achieves open-source SOTA performance, delivering greater stability in complex system engineering and deep debugging.

Improved real-world development experience: more reliable project-level context handling, adherence to engineering standards, and multi-platform development.

Support for 198K sequence length, with maximum input of 192K, maximum output of 128K, and maximum thought chain length of 64K.

Support for advanced functions, including thought chains, function calling, and prefix continuation.

GLM-5.1, the latest flagship model from Zhipu, delivers enhanced coding capabilities and superior performance on long-horizon tasks. It can autonomously operate for up to eight hours on a single task, covering the full cycle from planning and execution to iterative optimization, and producing engineering-level results.

While GLM-5.1 matches Claude Opus 4.6 in general intelligence and raw coding proficiency, it significantly outperforms the global frontier in long-horizon sustained execution. It excels at autonomous, multi-stage tasks over extended periods, making it the ideal foundation for building highly resilient autonomous agents and long-horizon coding engines.

It supports sequence lengths up to 198K, with a maximum input of 192K, a maximum output of 128K, and a maximum thought chain of 96K.

It supports chain-of-thought, function calling, and prefix continuation.

Compared with GLM-4.5, GLM-5 expands total parameters from 355B to 744B (boosting active parameters from 32B to 40B) and increases pre-training data from 23T to 28.5T tokens. In addition, GLM-5 integrates the DeepSeek sparse attention (DSA) mechanism, which significantly reduces deployment costs while maintaining the long context capability.

Compared to GLM-4.7, GLM-5 delivers substantial performance gains across a wide spectrum of academic benchmarks. It establishes a new state-of-the-art (SOTA) among global open-source models in reasoning, coding, and agentic tasks, further closing the capability gap with frontier closed-source models.

It supports sequence lengths up to 198K, with a maximum input of 192K, a maximum output of 64K, and a maximum thought chain of 64K.

It supports chain-of-thought, function calling, and prefix continuation.

Deep Thinking

For tutorials, see Deep Thinking. For API documentation, see Sending a Chat Request (Chat/Post).

**Table 1** Recommended models
Model Name	Version	model Parameter	Capability	Context Length	Max Input Length	Max Output Length	Max Chain-of-Thought Length	Supported Region
GLM-5.2	20260617	glm-5.2	Deep thinking (configurable) Function calling Prefix continuation	198K	192K	128K	64K	CN-Hong Kong
GLM-5.1	20260407	glm-5.1	Deep thinking (configurable) Function calling Prefix continuation	198K	192K	128K	96K	CN-Hong Kong
GLM-5	20260212	glm-5	Deep thinking (configurable) Function calling Prefix continuation	198K	192K	64K	64K	CN-Hong Kong
DeepSeek-V4-Pro	20260424	deepseek-v4-pro	Deep thinking (configurable) Function calling Prefix continuation	1M	1M	128K	96K	CN-Hong Kong
DeepSeek-V4-Flash	20260424	deepseek-v4-flash	Deep thinking (configurable) Function calling Prefix continuation	1M	1M	128K	96K	CN-Hong Kong
DeepSeek-V3.2	20251215	deepseek-v3.2	Deep thinking (configurable) Function calling Prefix continuation	160K	128K	32K	32K	CN-Hong Kong
DeepSeek-V3.1	20251124	deepseek-v3.1-terminus	Deep thinking (configurable) Function calling Prefix continuation	128K	96K	32K	32K	CN-Hong Kong

Text Generation

For tutorials, see Text Generation Overview. For API documentation, see Sending a Chat Request (Chat/Post).

**Table 2** Recommended models
Model Name	Version	model Parameter	Capability	Context Length	Max Input Length	Max Output Length	Max Chain-of-Thought Length	Supported Region
GLM-5.2	20260617	glm-5.2	Deep thinking (configurable) Function calling Prefix continuation	198K	192K	128K	64K	CN-Hong Kong
GLM-5.1	20260407	glm-5.1	Deep thinking (configurable) Function calling Prefix continuation	198K	192K	128K	96K	CN-Hong Kong
GLM-5	20260212	glm-5	Deep thinking (configurable) Function calling Prefix continuation	198K	192K	64K	64K	CN-Hong Kong
DeepSeek-V4-Pro	20260424	deepseek-v4-pro	Deep thinking (configurable) Function calling Prefix continuation	1M	1M	128K	96K	CN-Hong Kong
DeepSeek-V4-Flash	20260424	deepseek-v4-flash	Deep thinking (configurable) Function calling Prefix continuation	1M	1M	128K	96K	CN-Hong Kong
DeepSeek-V3	20250929	DeepSeek-V3	Function calling	128K	128K	32K	N/A	CN-Hong Kong
DeepSeek-R1-0528	20250929	deepseek-r1-250528	Deep thinking Function calling Prefix continuation	128K	96K	32K	32K	CN-Hong Kong

Next topic: Model List

Feedback

Was this page helpful?

Helpful Not helpful

Provide feedback

Thank you very much for your feedback. We will continue working to improve the documentation.See the reply and handling status in My Cloud VOC.

The system is busy. Please try again later.

For any further questions, feel free to contact us through the chatbot.

Chatbot