Summary & Key Takeaways ​

• Z.ai released GLM-5.2, a 753B parameter open-weights LLM under an MIT license.
• It features a massive 1 million token context window, a significant increase from its predecessor.
• GLM-5.2 leads the Artificial Analysis Intelligence Index for open-weights models.
• It ranks second on the Code Arena WebDev leaderboard, despite being text-only.
• The model is available via OpenRouter at competitive pricing compared to proprietary models.
• It uses more output tokens per task than other leading open-weights models.

Our Commentary ​

Okay, this is huge. A 753B parameter model with a 1 million token context window, open weights, and it's topping benchmarks? That's just wild. I'm genuinely impressed it's ranking so high on the Code Arena WebDev leaderboard without image input; I thought that was a prerequisite for good frontend coding. The token hunger is a bit of a bummer, but the sheer power and openness of this model are game-changers. We're seeing the open-source AI community push boundaries at an incredible pace. I'm excited to see what people build with this.