Select Language:
IT之家 reports that Zhipu has announced the launch of its new developer-exclusive package called the “GLM Coding Plan” for Claude Code, effective from September 1. This monthly subscription ranges from $20 to $100 and applies to their latest flagship models, GLM-4.5 and GLM-4.5-Air.
Initially, this package is centered around Claude Code, with plans to expand to incorporate additional coding tools in the future. The company confirmed that the GLM-4.5 model has already integrated several popular programming tools, including Claude Code, Cline, Gemini CLI, Grok CLI, CodeGeeX, Kilo Code, Roo Code, and Trae.
Zhipu emphasizes that the GLM-4.5 demonstrates consistent and practical performance across typical front-end development, cross-file modifications, and full-stack project building. User feedback suggests that its problem-solving capabilities are nearing those of Claude Sonnet 4. Furthermore, in multiple open-source evaluations, the GLM-4.5 maintains competitive standing alongside leading global models.
In WebDev Arena, it ranks as the second-best model alongside industry giants, while in SWE-bench Verified benchmarks, it outperforms Gemini-2.5-Pro and GPT-4.1. Notably, in the BFCL v4 benchmark, GLM-4.5 holds the top spot worldwide, surpassing Claude Opus 4.1 with a cost that is only 1.5% of the latter.
To assess its agentic coding capabilities, Zhipu developed the CC-Bench evaluation system, comparing GLM-4.5 against models such as Claude Sonnet 4, DeepSeek-V3.1, Kimi-K2, and Qwen3-Coder-480B across 52 real-world programming tasks spanning six key development areas. Results indicate that GLM-4.5 performs admirably among open-source models, especially in task execution quality and tool utilization reliability. Its coding performance closely rivals Claude Sonnet 4, yet at a fraction of the cost—just one-seventh of the latter’s price.
ChatGPT
Perplexity
Gemini AI
Grok AI
