GLM-5.2 reviewed: 1M token context, open weights, strong coding performance
AICodeKing received early access to Z AI's GLM-5.2, a post-trained upgrade to GLM-5.1 featuring a 1 million token context window, expected open weights with MIT licensing, and a benchmark score of ~81.43 — roughly 6% below Opus 4.8 and Fable.
Score breakdown
GLM-5.2's combination of a 1 million token context window, expected MIT-licensed open weights, and ~$8/month pricing places a near-frontier coding model within reach of developers who cannot afford or prefer not to use Claude or Codex pricing tiers.
- 01GLM-5.2 is a post-trained improvement over GLM-5.1, developed by Z AI
- 02Supports a 1 million token context window, described as a widely requested upgrade
- 03Expected to be released as open weights, likely under an MIT license
AICodeKing received early access to GLM-5.2 from Z AI ahead of its public launch. The model is a post-trained improvement over GLM-5.1, and Z AI has stated they hope it competes with Opus 4.8. A key upgrade is the addition of a 1 million token context window, which the video describes as a widely requested feature. GLM-5.2 is also expected to be released as open weights, likely under an MIT license, making self-hosting more accessible. It will be available through the GLM Coding Plan, which is compatible with Claude Code, Codex, OpenCode, and similar coding workflows. AICodeKing notes a preference for using it with OpenCode specifically.
The model is described as feeling faster, more focused, and more token-efficient than earlier versions.
On AICodeKing's own benchmark, GLM-5.2 scored approximately 81.43 — about 6% below both Opus 4.8 and Fable. Task-by-task results from the review include: an elevator simulator (8/10, correct behavior with minor alignment issues), a Three.js 3D contact lens case (3/10, poor proportions), a Three.js folding table with a slider (9/10, described as better than Fable on this task), an SVG panda eating a burger (8/10, praised for detail and character), and a bow and arrow simulator game (details truncated in source). The model is described as feeling faster, more focused, and more token-efficient than earlier versions. The GLM Coding Plan is priced at around $8 per month, which the video frames as strong value relative to Claude and Codex pricing.
Key facts
- 01GLM-5.2 is a post-trained improvement over GLM-5.1, developed by Z AI
- 02Supports a 1 million token context window, described as a widely requested upgrade
- 03Expected to be released as open weights, likely under an MIT license
- 04Scored ~81.43 on AICodeKing's personal benchmark, approximately 6% below Opus 4.8 and Fable
- 05Available via the GLM Coding Plan, compatible with Claude Code, Codex, and OpenCode
- 06GLM Coding Plan is priced at around $8 per month
- 07Task scores ranged from 3/10 (Three.js contact lens case) to 9/10 (Three.js folding table)
Topics
Summary and scoring are generated automatically from the original article. We always link back to the publisher and never republish images or paywalled content. Last processed Jun 14, 2026 · 09:08 UTC. How this works →