Zhipu AI drops GLM-5.2 coding model

// 1h agoMODEL RELEASE

Zhipu AI drops GLM-5.2 coding model

ANNOUNCEMENT PRODUCT GITHUB PRODUCT HUNT

Zhipu AI has launched GLM-5.2, its new 744-billion parameter Mixture-of-Experts coding model featuring a 1-million-token context window. Currently available to GLM Coding Plan subscribers, Zhipu AI plans to release the model's MIT-licensed weights and API next week.

// ANALYSIS

Zhipu AI's release of GLM-5.2 directly challenges proprietary frontier models by offering a massive context window and strong coding reasoning.

–The 1-million-token context window is highly suited for codebase-wide refactoring and long-horizon developer agents.
–The upcoming MIT-licensed open weights release provides a powerful, customizable alternative to commercial APIs.
–Immediate access is restricted to paying subscribers, meaning third-party benchmarks and community verification are not yet available.

// TAGS

glm-5.2zhipu-aillmmoeopen-weightslong-contextai-codingagentopen-source

DISCOVERED

1h ago

2026-06-13

PUBLISHED

1h ago

2026-06-13

RELEVANCE

8/ 10

AUTHOR

Income stream surfers

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

UPDATE39m ago

OpenCode 1.17.6 boosts agent stability

OpenCode version 1.17.6 has been released, bringing stability updates to the terminal-native AI coding agent. The update skips unavailable copy roots during project refreshes, improves feedback when moving sessions, and implements explicit client capability negotiations for Model Context Protocol (MCP) integration.

VIDEO1h ago

Nous Research showcased their Hermes Agent autonomously learning and controlling TouchDesigner from scratch to build creative visual systems.

A demonstration shared by creator Amber Shen highlights the capabilities of Nous Research's Hermes Agent, which was able to autonomously learn and build in the node-based visual programming environment TouchDesigner. Operating via the Model Context Protocol (MCP), the local AI agent navigated the desktop environment to construct interactive real-time 3D visuals without manual user intervention.

MODEL1h ago

Claude Fable 5 exposes fragile safety guardrails

Anthropic's Claude Fable 5 was introduced as a safeguarded, public-facing variant of its powerful Mythos-class architecture designed for complex agentic workflows. The release strategy relies on real-time classifiers that reroute sensitive prompts to Claude Opus 4.8, depending on the premise that software guardrails can successfully isolate hazardous capabilities.

Zhipu AI drops GLM-5.2 coding model