AIToday
Large Language ModelsOpen-Source AIAI Business & IndustryFortune AIPublished: Aug 28, 2026, 04:00 JST2 min read

Chinese open-source AI gains ground with U.S. businesses

Chinese open-source AI gains ground with U.S. businesses

Key takeaway

  • Chinese open-source AI models are increasingly being adopted by U.S. businesses. Spending on model serving platforms rose to 6.1% in July.

  • These models offer lower cost and more control.

  • The shift may pressure U.S. labs on pricing.

3 Key Points

  1. What happened

    A growing share of businesses are paying for model serving platforms that offer open-source and Chinese-developed models. This share rose to 6.1% of AI-spending businesses in July, up from 4.5% in January 2026, according to Ramp's AI Index.

  2. Why it matters

    Open-weight models are attractive because they offer more fine-tuning opportunities, are much cheaper, and give enterprises more control, including assurance that their data isn't used to train competing products. Notable examples include Moonshot's Kimi K3 and Z.AI's GLM-5.3-Flash.

  3. What to watch

    Whether U.S. frontier labs have hit a price ceiling. Anthropic's Fable 5, priced at roughly $10 per million tokens, accounts for just 6% of Anthropic's tokens and 11.4% of spend, while cheaper alternatives are gaining traction.

Ask the AI about this article →

Context & Analysis

Chinese open-weight AI models are increasingly seen as viable alternatives to closed-source U.S. models. Ramp's data shows a rise in businesses using model serving platforms that host such models, indicating growing interest. This is partly because open-weight models offer cost savings and customization, as noted by Alex Brunicki of Backed VC. Companies like Thomson Reuters and Harvey have already adopted open-source models for specific tasks, citing lower cost and adequate performance.

The pricing pressure is evident. Anthropic's Fable 5, despite being advanced, captures only a small share of Anthropic's token and dollar spend, suggesting businesses are not willing to pay a premium when cheaper models suffice. This trend may force U.S. labs to reconsider their pricing strategies. The potential release of a larger GLM-5.3 model with strong cyber capabilities could further disrupt the market, though its impact remains to be seen.

FAQ

What are open-weight AI models?
Open-weight models are AI systems whose underlying parameters are publicly available, allowing companies to fine-tune them for specific tasks. They are often cheaper and offer more control than closed-source models.
Which Chinese models are mentioned?
The article highlights Moonshot's Kimi K3 and Z.AI's GLM-5.3-Flash (formerly Ox Alpha). Kimi K3 is noted for its large size and performance close to leading proprietary systems.
How much does GLM-5.3-Flash cost?
Z.AI will charge $0.15 per million input tokens and $0.50 per million output tokens.

Get the latest Large Language Models news every morning

AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.

Free · takes 30 seconds · unsubscribe anytime

Ask AI

Ask AI anything about this article. Q&As are published on this page for other readers too.

Related Articles

Next articleAnthropic unveils Model Hardware Standard for safe AI agents in labs and factories