# Z.ai's open-weights GLM-5.2 beats GPT-5.5 on multiple long-horizon coding benchmarks for 1/6th the cost

_Thursday, July 9, 2026 at 8:00 AM EDT · AI · Latest · Tier 2 — Notable_

![Z.ai's open-weights GLM-5.2 beats GPT-5.5 on multiple long-horizon coding benchmarks for 1/6th the cost — Primary](https://images.ctfassets.net/jdtwqhzvc2n1/2wFQ5P7x9M73wwoWGI1RyM/12f4b572d351a53f3d4a3741de6b3839/Gemini_Generated_Image_e0yu6he0yu6he0yu.png?w=800&q=75)

Chinese AI startup Z.ai announced the immediate release of GLM-5.2, a 753-billion parameter open-weights large language model engineered for long-horizon autonomous coding tasks. The company said the model is available on Hugging Face, the Z.ai API, and more than 20 third-party coding environments with a 1-million-token context window. Z.ai released the core weights under an unrestricted MIT open-source license, allowing enterprises to download, customize, and run the model locally. The company said the model introduces an architectural optimization called IndexShare, which reuses an indexer across every four sparse attention layers and reduces per-token compute FLOPs by 2.9 times at maximum context length. GLM-5.2 also features an upgraded Multi-Token Prediction layer and selectable thinking modes. On industry-standard benchmarks, Z.ai said GLM-5.2 scored 62.1 on SWE-bench Pro, beating GPT-5.5 at 58.6. On FrontierSWE, GLM-5.2 hit 74.4%, surpassing GPT-5.5 at 72.6%. On MCP-Atlas, GLM-5.2 achieved 77.0, outscoring GPT-5.5 at 75.3. On Humanity's Last Exam with tools, GLM-5.2 reached 54.7, ahead of GPT-5.5 at 52.2. Z.ai launched the GLM Coding Plan with tiers starting at $12.60 per month. API pricing is $1.40 per million input tokens and $4.40 per million output tokens.

## Sources

- [venturebeat.com](https://venturebeat.com/technology/z-ais-open-weights-glm-5-2-beats-gpt-5-5-on-multiple-long-horizon-coding-benchmarks-for-1-6th-the-cost)

---
Canonical: https://techandbusiness.org/newswire/1zexheHxKDYI99qGZra13D
Retrieved: 2026-07-10T03:15:53.712Z
Publisher: Tech & Business (techandbusiness.org)