Article · 2026-06-22 · 6 moments

GLM-5.2 is the step change for open agents

A capability threshold I've been carefully monitoring. ✦ AI generated

Claim◆

Banning open models now while closed models race 10-100X ahead in the hands of one or two companies creates a bigger long-term risk than allowing open Mythos-class models to exist.

Despite acknowledging the fear of an openly accessible frontier-class model, the author argues that suppressing open models while closed labs monopolize massive capability gains would create a worse outcome.

transcript

Nathan Lambert: 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.

gives example · 1

Anecdote

GLM-5.2 is the first open-weight model that feels right as a general coding agent in real harness use, despite minor integration bugs like image inputs bricking a Fireworks API session.

Testing GLM-5.2 hands-on via Fireworks' API inside Claude Code to build course content, the author found setup easy and capabilities immediately convincing, aside from a bug where sending images would brick the API session.

transcript

Nathan Lambert: I put it to work helping make content for my post-training course with Fireworks' API in Claude Code (setting this up was very easy). There were some minor knife cuts, such as the Claude Code harness / my repo documentation trying to send images to the model, which would brick Fireworks API for the session — forcing a manual context clear.

supports · 1

Prediction

GLM-5.2's diffusion while Claude Fable remains export-banned deals a severe economic blow to frontier labs, letting open models erode the lower-margin business those labs need while they push into higher-margin frontier domains.

The author predicts serious pricing pressure and revenue risk for Anthropic as GLM-5.2 spreads through the market while Anthropic's flagship model is barred from sale, calling it an 'economic dagger.'

transcript

Nathan Lambert: This diffusion happening while Anthropic's, and by extension the U.S.'s flagship model, is still banned is a severe economic dagger. GLM-5.2 is being given time to carve out the economic underbelly of the frontier labs when they want to be pushing forward into higher margin, higher revenue domains enabled only by the absolute frontier models.

provides context · 1

Claim◆

GLM-5.2's release has generated a community focal point rivaled only by DeepSeek R1, exceeding even the 'DeepSeek Moment' the author previously used to describe Kimi K2's release.

The author argues GLM-5.2 is only the second open-model release (after DeepSeek R1) to become such a dominant topic of discussion, surpassing his own earlier 'DeepSeek Moment' comparison for Kimi K2.

transcript

Nathan Lambert: Such a focal point of discussion among the community has only been so clear with an open model release once before — DeepSeek R1. This is not a comparison I make lightly, and when I compared Kimi K2's release to a "DeepSeek Moment," GLM-5.2 has well exceeded that.

supports · 1

Data

The 204-day gap between Claude Opus 4.5's release and GLM-5.2's release confirms the commonly cited 6-9 month performance lag between U.S. closed labs and Chinese open-weight labs.

Calculating the exact time gap between Opus 4.5 and GLM-5.2 (204 days, ~6.8 months), the author finds it lines up precisely with the widely claimed 6-9 month open-closed capability lag.

transcript

Nathan Lambert: With Claude Opus 4.5's release on November 24th, 2025, the gap in time to GLM-5.2's release on June 16th, 2026 is 204 days — or about 6.8 months. This puts us square in the 6-9 month time gap that many people claim as the performance lag between the U.S.'s closed labs and China's open counterparts.

explains mechanism · 1

Context

Z.ai timed the GLM-5.2 release to capitalize on the backlash against Anthropic's export restriction and effective banning of Claude Fable 5, continuing a pattern of Chinese open-weight labs seizing easy marketing wins.

The author frames GLM-5.2's unusual Saturday rollout as Z.ai exploiting the PR fallout from Anthropic's Claude Fable 5 export ban, part of a recurring pattern of Chinese labs grabbing marketing opportunities.

transcript

Nathan Lambert: In this case, it seemed like Z.ai was excited to capitalize on the zeitgeist of "Anthropic being anti open-science" with their silent safeguards on AI researchers. For the past year or two, the Chinese open-weight labs have taken every opportunity they have for easy marketing wins like this.

Highlight slides

GLM-5.2's Release: A New Focal Point✦ from: GLM-5.2's release has generated a community focal point rivaled only by DeepSeek R1, exceeding even the 'DeepSeek Moment' the author previously used to describe Kimi K2's release.The Real Risk Isn't Open Models✦ from: Banning open models now while closed models race 10-100X ahead in the hands of one or two companies creates a bigger long-term risk than allowing open Mythos-class models to exist.Only the Second Such Moment✦ from: GLM-5.2's release has generated a community focal point rivaled only by DeepSeek R1, exceeding even the 'DeepSeek Moment' the author previously used to describe Kimi K2's release.Projected Closed-Model Capability Gains (2 Years)✦ from: Banning open models now while closed models race 10-100X ahead in the hands of one or two companies creates a bigger long-term risk than allowing open Mythos-class models to exist.

Related episodes

Article · Simon Willison's Weblog

Banning open models now while closed models race 10-100X ahead in the hands of one or two companies creates a bigger long-term risk than allowing open Mythos-class models to exist.

GLM-5.2 is the first open-weight model that feels right as a general coding agent in real harness use, despite minor integration bugs like image inputs bricking a Fireworks API session.

GLM-5.2's diffusion while Claude Fable remains export-banned deals a severe economic blow to frontier labs, letting open models erode the lower-margin business those labs need while they push into higher-margin frontier domains.

GLM-5.2's release has generated a community focal point rivaled only by DeepSeek R1, exceeding even the 'DeepSeek Moment' the author previously used to describe Kimi K2's release.

The 204-day gap between Claude Opus 4.5's release and GLM-5.2's release confirms the commonly cited 6-9 month performance lag between U.S. closed labs and Chinese open-weight labs.

Z.ai timed the GLM-5.2 release to capitalize on the backlash against Anthropic's export restriction and effective banning of Claude Fable 5, continuing a pattern of Chinese open-weight labs seizing easy marketing wins.

Fable's judgement

Slow down to speed up: so much has changed in 6 months’ time

[AINews] OpenAI GPT-5.6 Sol / Terra / Luna — restricted to trusted partners