Z.ai’s Open-Weight GLM-5.2 Surpasses GPT-5.5 in Long-Horizon Coding While Cutting Costs by 83%

Open-weight AI just had its biggest week yet. On June 16, 2026, Z.ai, the Chinese AI startup previously known as Zhipu AI, unveiled GLM-5.2, a 753-billion-parameter open-weight model designed to handle complex coding and long-duration engineering workflows. GLM-5.2 outperforms GPT-5.5 on multiple coding benchmarks while costing roughly one-sixth as much to operate. The model is available immediately on Hugging Face, the Z.ai API, and more than 20 third-party coding environments, with a stable 1-million-token context window. That combination of price and performance is forcing a hard look at the open-versus-closed AI debate. 

The Benchmark Numbers That Made Headlines

On SWE-bench Pro, which tests real-world software engineering tasks, GLM-5.2 scored 62.1, decisively outperforming GPT-5.5’s 58.6. That margin alone reset expectations for open-weight models.

• On FrontierSWE, designed to measure long-horizon task completion, GLM-5.2 hit 74.4%, beating GPT-5.5’s 72.6%.
• GLM-5.2 came close to Anthropic’s Claude Opus 4.8, which scored 75.1% on the same FrontierSWE benchmark.
• On MCP-Atlas, a tool-usage benchmark, GLM-5.2 scored 77.0 against GPT-5.5’s 75.3.
• On Humanity’s Last Exam with tools enabled, GLM-5.2 scored 54.7, ahead of GPT-5.5’s 52.2.
• GLM-5.2 also posted 81.0 on Terminal-Bench 2.1, the strongest score among all open-source models on that test.

What “Open-Weight” Actually Means Here

MIT License, No Restrictions

The full weights are live on Hugging Face under the handle zai-org/GLM-5.2, released under an MIT open-source license. That license carries real commercial weight for developers. Z.ai’s technical documentation explicitly states the license guarantees “no regional limits” and allows “technical access without borders.” Enterprises can download the model freely, customize or fine-tune it, and run it locally for only the cost of compute and electricity. That structure removes the licensing friction that closed models like GPT-5.5 carry by default.

The Pricing Gap Driving Enterprise Interest

GPT-5.5 charges $5 per million input tokens and $30 per million output tokens, available only through OpenAI’s closed API. That pricing structure is exactly what GLM-5.2 is built to undercut.

• Z.ai’s enterprise subscription tiers start at just $12.60 per month.
• GLM-5.2 is live now across every GLM Coding Plan tier: Lite, Pro, Max, and Team.
• The model supports up to 131,072 output tokens per response, roughly five times larger than GLM-5.1’s 200,000-token window.
• GLM-5.2 builds on the same 744-billion-parameter Mixture-of-Experts architecture as GLM-5, adding a new dual thinking-effort system labeled High and Max.

What This Means for the Broader AI Race

GLM-5.2 is not Z.ai’s first challenge to Western flagship models. GLM-5, two iterations earlier, already beat GPT-5.2 on several coding benchmarks when it launched. The pattern is now a trend, not a one-off result.

In some blind tests conducted by humans, GLM-5.2 has even outperformed Claude Fable 5, according to reporting from Z.ai. Markets appear to interpret this development as a sign of rising competition in the AI model landscape, potentially affecting the standing of other major models. Companies with direct exposure to this competitive shift include Microsoft (NASDAQ: MSFT), Alphabet (NASDAQ: GOOGL), and Nvidia (NASDAQ: NVDA), all tied closely to AI model demand and the compute infrastructure behind it.

Final Thoughts

GLM-5.2 marks the third major release in Z.ai’s GLM-5 family within months, a release cadence few AI labs have matched. For sovereign-data, regulated, or air-gapped enterprise requirements, GLM-5.2’s self-hosted option is currently the only credible choice among frontier-tier coding models. Whether GPT-5.5 responds with a price cut or a performance update will likely arrive within weeks, given how fast this rivalry has moved in 2026. For now, GLM-5.2 holds a rare position: faster, cheaper, and open, all at once.

Disclaimer

The content shared by Meyka AI PTY LTD is solely for research and informational purposes. Meyka is not a financial advisory service, and the information provided should not be considered investment or trading advice.