Chinese AI startup Z.ai released GLM-5.1, a large language model designed to perform autonomous work for extended periods without human intervention. The 754-billion parameter model is available under the MIT License, allowing commercial use and modification.

The company claims GLM-5.1 can maintain focus on complex tasks for up to eight hours, significantly longer than previous models that typically plateau after 20-50 steps. The model demonstrated this capability in benchmark tests, including optimizing a vector database where it performed over 6,000 tool calls across 655 iterations to achieve performance improvements of roughly six times baseline results.

On coding benchmarks, GLM-5.1 scored 58.4 on SWE-Bench Pro, outperforming GPT-5.4 (57.7), Claude Opus 4.6 (57.3), and Gemini 3.1 Pro (54.2) in resolving real-world software engineering issues. The model also showed strong performance on mathematical reasoning tasks, scoring 95.3 on the AIME 2026 competition benchmark.

Z.ai has structured its offerings with GLM-5.1 as the open-source option while keeping GLM-5 Turbo as a proprietary, faster variant. The company offers subscription tiers ranging from $27 to $216 per quarter, with API pricing at $1.40 per million input tokens and $4.40 per million output tokens.

The release comes as the AI industry increasingly focuses on models capable of sustained autonomous operation rather than just quick responses. Z.ai, which went public on the Hong Kong Stock Exchange with a market cap of $52.83 billion, is positioning itself as a leader in what it calls "agentic engineering" - AI systems that can complete entire projects with minimal supervision.