GLM-5.2 is described by Alibaba Cloud as the latest flagship model from Zhipu AI for long-horizon tasks, with support for an ultra-long 1M context window and strong logical reasoning, long-text comprehension, and code generation.
Alibaba Cloud also states that GLM-5.2 is a text-to-text model and that it can be accessed through OpenAI-compatible, DashScope, and Anthropic-compatible interfaces.
| Developer | Zhipu AI |
|---|---|
| Model ID | alibaba/glm-5.2 |
| Release Date | June 15, 2026 |
| Context Window | 1,000,000 tokens (~1M) |
| Max Output Tokens | 131,072 tokens |
| Modalities | Text → Text |
| Capabilities | parallel tool calls, reasoning, streaming, structured (JSON schema) output, tool/function calling |
Long-horizon focus
The model is positioned for long-horizon tasks and is described as suitable for situations that require sustained reasoning over long inputs.
1M context window
Alibaba Cloud documents GLM-5.2 as supporting a 1M context window, and the model overview calls it the latest GLM model with a context length of 1M.
Reasoning and code generation
The model is described as having powerful logical reasoning, long-text comprehension, and code generation capabilities.
Thinking mode
Alibaba Cloud says GLM-5.2 has thinking mode enabled by default, returns `reasoning_content` before the final answer, and supports the `reasoning_effort` parameter.
Tool and structured-output support
The model supports Function Calling and structured output, and Alibaba Cloud notes that structured output is supported only in non-thinking mode.
Complex enterprise workflows
Alibaba Cloud describes GLM-5.2 as suitable for intelligent interaction and enterprise applications, which aligns with its long-horizon reasoning and long-text handling focus.
Developer assistance
The model is presented as useful for development assistance, especially for code generation and multi-step reasoning tasks.
Agentic tool use
GLM-5.2 supports Function Calling and, through the Responses API, can use tools such as web search, web extractor, and code interpreter.
Streaming interactions
Alibaba Cloud documents support for streaming and the `tool_stream` parameter, which incrementally returns tool-call arguments during streamed generation when enabled.
Sources: https://www.alibabacloud.com/help/en/model-studio/glm-5-2 · https://www.alibabacloud.com/help/en/model-studio/glm-5-2-us · https://www.alibabacloud.com/help/en/model-studio/glm
Measured on the AI/ML API gateway across live traffic. Figures are medians unless stated. Last updated 2026-10-12.
GLM 5.2 is documented as suited to AI applications and copilots, content creation, workflow automation, and video generation pipelines. It is also described as well-suited for agentic coding, tool use, reasoning, and long-context tasks. Alibaba Cloud’s documentation says it is suitable for intelligent interaction, enterprise applications, and development assistance scenarios. NVIDIA’s catalog describes it as a fit for AI agent systems, chatbots, and applications requiring advanced reasoning and long-context understanding.
import requests r = requests.post( "https://api.aimlapi.com/v1/chat/completions", headers={"Authorization": "Bearer " + AIMLAPI_KEY}, json={ "model": "alibaba/glm-5.2", "messages": [ { "role": "user", "content": "Hello!" } ] }, ) print(r.json())
const r = await fetch("https://api.aimlapi.com/v1/chat/completions", { method: "POST", headers: { Authorization: `Bearer ${process.env.AIMLAPI_KEY}`, "Content-Type": "application/json", }, body: JSON.stringify({ "model": "alibaba/glm-5.2", "messages": [ { "role": "user", "content": "Hello!" } ] }), }); console.log(await r.json());
curl -X POST https://api.aimlapi.com/v1/chat/completions \ -H "Authorization: Bearer $AIMLAPI_KEY" \ -H "Content-Type: application/json" \ -d '{"model":"alibaba/glm-5.2","messages":[{"role":"user","content":"Hello!"}]}'
OpenAI-compatible — swap the base URL and it works with your existing SDK.
| Type | Price |
|---|---|
| Input | |
| Output | |
| Cached input |
GLM-5.2’s documented pricing caveat is that cached input is billed at a discounted rate of $0.26 per million tokens, while Z.ai’s pricing also notes limited-time free cached-input storage.
| Benchmark | Score | What it measures | Source | Retrieved |
|---|---|---|---|---|
| Intelligence | 34 | Composite score across standardised reasoning, knowledge and problem-solving evaluations, measured independently by Artificial Analysis | Source | September 12, 2026 |
| Coding | 68.8 | Composite score across standardised coding evaluations, measured independently by Artificial Analysis | Source | September 12, 2026 |
| GPQA Diamond | 91.2 | Google-proof graduate science questions (hardest subset) | Source | September 21, 2026 |
| Humanity's Last Exam | 40.5 | Expert-level questions across many domains | Source | September 21, 2026 |
| Model | Input | Output | Context | Best for |
|---|---|---|---|---|
GLM 5.2 This page | Agentic workflows and structured output |
GLM 5.2 has a 1,000,000 tokens context window and can return up to 131,072 tokens.
GLM 5.2 takes text as input and returns text.
Use alibaba/glm-5.2 as the model id. Requests go to https://api.aimlapi.com/v1/chat/completions.
GLM 5.2 became available on June 16, 2026.
GLM 5.2 is priced at input $1.82 / 1M tokens, output $5.72 / 1M tokens, cached input $0.364 / 1M tokens.
Yes, GLM 5.2 can stream responses as they are generated.
GLM 5.2 was built by Zhipu AI.
