China’s Zhipu AI (智谱AI) open-sources GLM-5.1, raises prices to narrow gap with US rivals

Open-source, but pricier

Zhipu AI (智谱AI) has open-sourced its latest flagship model, GLM-5.1, while simultaneously raising API prices by about 10 per cent — its second price hike this year. The move, reported by the South China Morning Post, is aimed at monetising advanced AI capabilities as competition with US firms intensifies. Zhipu’s Hong Kong-listed shares jumped roughly 11.5 per cent on the announcement, a sign that investors welcomed the revenue-first tilt.

Performance and pricing comparisons

GLM-5.1 has been ranked by benchmarking firm Artificial Analysis as the strongest open model globally on overall intelligence, outperforming several domestic rivals such as MiniMax, but still trailing leading US systems from OpenAI and Anthropic. Zhipu now charges roughly US$1.40 per million input tokens and US$4.40 per million output tokens. By comparison, Anthropic’s Claude Opus 4.6 was priced at about US$5 per million input and US$25 per million output as of February 2026, leaving a meaningful — if narrower — gap.

Strategy: ecosystem building meets monetisation

The combination of open-sourcing technology while lifting prices reflects a dual strategy: cultivate a developer ecosystem and academic credibility through openness, while extracting more value from commercial users. It has been reported that GLM-5.1 was initially released as a proprietary feature for coding subscribers late last month, feeding speculation that Zhipu is rebalancing away from a pure open-source model towards revenue generation. Open-sourcing the weights can help drive adoption and third-party innovation — but charging more for cloud tokens captures the commercial upside.

Geopolitics and what comes next

The shift also comes against a backdrop of US–China tech rivalry, export controls on advanced chips and growing regulatory scrutiny that shape how Chinese AI firms scale and monetise products. Will more Chinese labs follow Zhipu’s hybrid playbook — open research, paid inference? If so, expect sharper competition on price-performance and clearer commercial tiering between open and paid offerings as Beijing pushes for self-reliance in AI.