Z.ai GLM 5.2
Z.aiFlagship
ThinkingTool UseStructured Output
About this model
GLM 5.2 is a large-scale reasoning model from Z.ai. It supports text input and output with a 1M-token context window, and is suited for long-horizon agent workflows, project-level software engineering,...
Performance Tier
Flagship
Z.ai GLM 5.2 is a flagship model from Z.ai : the most capable in their lineup.
Best-in-class model from this provider. Highest performance across benchmarks, ideal for demanding tasks.
Pricing
This model is included in Elosia plansModerate
Moderate cost. A balanced choice for regular use without constant cap watching.
| Type | per 1M tokens |
|---|---|
| Input (prompt) | $0.980 |
| Output (completion) | $3.08 |
| Cache read | $0.182 |
Capabilities
Context Length1.0M
Max Output Tokens—
TokenizerOther
Inputtext
Outputtext
Release DateJune 16, 2026
Benchmarks
General Intelligence
MMLU
Not reported
GPQA Diamond
91.2%
Mathematics
MATH-500
Not reported
AIME 2026
99.2%
Programming
HumanEval
Not reported
SWE-bench Verified
Not reported
Reasoning
IFEval
Not reported
Agentic
SWE-bench Pro
62.1%
Recommended Use Cases
CodingAnalysisResearch
Strengths
- MoE architecture (~750B total, 40B active) — top open-weight model for long-horizon agentic coding (SWE-bench Pro 62.1, beating GPT-5.5)
- 1M-token context window (up to 128K output), made practical by IndexShare sparse attention — full-repository and book-length inputs in a single pass
- Frontier-competitive on hard reasoning — GPQA Diamond 91.2 and AIME 2026 99.2, on par with leading Western models
- Open-weight under MIT license — unrestricted commercial self-hosting, weights on HuggingFace and ModelScope
- Cost-efficient at $0.98/M input and $3.08/M output — a fraction of comparable frontier models for long-horizon coding
Limitations
- Text-only : no image or multimodal input
- Headline benchmarks are self-reported by Z.ai; independent verification remains limited
- Narrowly optimized for coding, agentic and reasoning tasks — weaker on broad conversational use (around #25 on Text Arena)
Resources
This model may use your data for training
