GLM-5.1
Positioning and release
GLM-5.1 is a flagship text model designed for agentic engineering and long-horizon tasks. It was announced on April 7, 2026. Zhipu AI released its model weights under the MIT License. (z.ai)
Capabilities and uses
- Supports text input and output, thinking modes, streaming, function calling, context caching, structured output and MCP integration. The documented context length is 200K tokens, with a maximum output of 128K tokens. (docs.z.ai)
- Intended for coding agents, complex software engineering, repository generation, general conversation, creative writing, front-end prototyping and document workflows. It is available through Z.AI and BigModel developer platforms; the announcement also describes local deployment using frameworks including vLLM and SGLang. (z.ai)
- Zhipu reports a 58.4 SWE-Bench Pro score and describes an experiment involving more than 600 optimization iterations and 6,000 tool calls. These are publisher-reported results under the stated evaluation conditions, not guarantees for every deployment. (z.ai)
Limitations
The documented model is text-only. Context and output limits constrain individual requests, and long-running agent tasks depend on the surrounding tools, infrastructure and permissions. Generated code and other outputs should be reviewed and tested; benchmark results do not guarantee correctness or autonomous task completion. (docs.z.ai)
API pricing
Prices below are platform-specific, not currency conversions. Both official pricing pages were checked on November 16, 2026.
BigModel China — CNY
Pay-as-you-go GLM-5.1 text-model prices, in CNY per 1 million tokens: (docs.bigmodel.cn)
| Input length | Input | Output | Cached input |
|---|---|---|---|
| [0, 32K) | 6 | 24 | 1.3 |
| ≥32K | 8 | 28 | 2 |
Cache storage is listed as free for a limited time. The tier is determined by input length; cached input is billed separately from uncached input. Official source: BigModel API pricing. (docs.bigmodel.cn)
Z.AI international — USD
The official GLM-5.1 text-model price is USD 1.4 per 1 million input tokens, USD 4.4 per 1 million output tokens, and USD 0.26 per 1 million cached input tokens. Cached-input storage is listed as free for a limited time. Official source: Z.AI pricing. (docs.z.ai)