GLM-5.2 is the step change for open agents
Interconnects · Nathan Lambert · 2026-06-22
Nathan Lambert argues that Z.ai's GLM-5.2, released June 2026 during U.S. export restrictions on Claude Fable 5, is the first open-weight model to credibly compete with frontier closed models in coding agent workflows, representing an inflection point comparable to DeepSeek R1's impact on reasoning capabilities.
Appears in
Extraction
Topics: open-weight-modelscoding-agentsai-competitionchina-aiai-policy
Claims
- GLM-5.2 is the first open-weight model to perform credibly as a general agent in coding harnesses like Claude Code, with Arena's agent leaderboard placing it alongside OpenAI and Anthropic's latest closed models.
- The performance gap between U.S. closed-model labs and Chinese open-weight labs is approximately 6-9 months, evidenced by GLM-5.2 reaching Opus 4.5 capability levels 204 days after Opus 4.5's November 2025 release.
- GLM-5.2's release during the export restriction on Claude Fable 5 creates significant economic pressure on Anthropic by offering open alternatives to Claude Code's dominant position in coding agent workflows.
- The trajectory of increasingly capable open-weight Chinese models raises the possibility of U.S. government attempts to restrict access to such models, which Lambert views as a dangerous concentration of AI power.
- Open-model inference providers including Fireworks, Together, and Prime Intellect hit a new commercial inflection point with GLM-5.2 as demand for frontier-quality open model inference grows.
Key quotes
GLM-5.2 is the open weight model that feels right in coding harnesses as a general agent. It's the first one.
Anthropic's record revenue growth rate on the back of Claude Code is heavily driven by being the best model, and the only model that can really do this. GLM-5.2 is the first of many (coming soon) open weight models to offer credible alternatives.
I totally see why it is scary to imagine an openly accessible Mythos class model, but if open models get banned now and only closed models get 10 or 100X better in 2 years in the hands of one or two companies, I think we will have bigger problems on our hands.