GLM-5.2

GLM-5.2 is positioned by its publisher as a flagship open-source large language model aimed at long-horizon coding, agentic workflows, and demanding reasoning tasks. The release was framed publicly as a step toward broadly accessible frontier intelligence, with the founder of Z.AI announcing that GLM-5.2 is fully open. That framing drew notable community interest, with a Hacker News thread titled "GLM 5.2 Is Out" reaching 772 points and drawing hundreds of comments, signaling strong developer engagement around the launch.

In practical terms, GLM-5.2 is intended for complex reasoning, advanced software engineering, and large-scale data processing scenarios where extended context and structured outputs matter. DeepInfra published an integration guide for the model, indicating third-party hosting and tooling support beyond Z.AI's own distribution. The combination of open-weight availability and a design focus on agentic and tool-driven tasks makes GLM-5.2 a practical fit for teams building long-running coding assistants, multi-step automation pipelines, and reasoning-heavy applications that benefit from an open-weights foundation.

GMI Cloudzai-org/GLM-5.2-FP8glm

Open provider page Provider docs

Quick Info

Provider: GMI Cloud
Model key: zai-org/GLM-5.2-FP8
Release date: Jun 13, 2026
Last updated: Jun 13, 2026
Input modalities
Output modalities
Capabilities

Cost

Input token cost: $0.979
Output token cost: $3.08

Limits

Output tokens: 131,072 tokens
Context window: 1,000,000 tokens

Transparent token rates

Compare GLM-5.2 pricing

Rates are shown per one million tokens. Combined means one million input plus one million output tokens.

Browse this family

Latest news about GLM-5.2

GMI Cloud

Official sourceAnnouncement

GMI Cloud published a first-party post announcing that GLM-5.2 is available on its platform, positioning it as a practical open-weight coding model. The post confirms a 1,000,000-token context window, an MIT license for the weights (available on Hugging Face), and a 744-billion-parameter Mixture-of-Experts architecture The article frames GLM-5.2 as optimized for long-horizon coding agent scenarios, including large-scale implementation, automated research, and performance optimization, with months of specialized training to make the 1M context usable rather than purely theoretical. GMI Cloud highlights the model's ability to load enti

GMI Cloud

CoverageAnalysis

Artificial Analysis reported on June 16, 2026, that GLM-5.2 is the new leading open-weights model on its Intelligence Index v4.1 with a score of 51, placing it ahead of MiniMax-M3 (44), DeepSeek V4 Pro max (44), and Kimi K2.6 (43). The model is the same size as its predecessor (744B total / 40B active parameters) but s GLM-5.2 shows notable gains on scientific reasoning (CritPt +16 points to 21%, HLE +12 points to 40%), AA-LCR (+9 points to 71%), tau3 banking (+15 points to 27%), and SciCode (+7 points to 50%), with TerminalBench v2.1 improving 16 points to 78% and GPQA Diamond reaching 89%. It scores 1524 on GDPval-AA v2, ahead of M

GMI Cloud

Coverage

DeepInfra announced on July 1, 2026 that GLM-5.2 is available on its platform as zai-org/GLM-5.2, with the headline feature being a stable 1,048,576-token (1M) context window designed for reliable long-horizon work rather than the degraded retrieval typical of million-token claims. The model ships under an MIT license DeepInfra details the architecture behind the long context: GLM-5.2 introduces IndexShare, which reuses the same indexer across every four sparse attention layers, cutting per-token FLOPs by 2.9x at 1M context length and making the long window practically affordable. On the inference side, the model ships with an impro