A recent Mozilla report indicates that China's open-weight AI models are rapidly closing the gap with leading US frontier models, trailing by an estimated 4.4 months in task completion time. While these models are significantly cheaper to use, they still lag in certain benchmarks and can be resource-intensive to run. Despite the narrowing capability gap, closed-source models continue to dominate revenue on platforms like OpenRouter. AI
IMPACT Suggests increased competition and potentially lower costs for AI model usage as open-weight models mature.
RANK_REASON Report on AI model capabilities and development trends.
- China
- Mozilla
- U.S.
- Claude Fable-5
- Claude Opus 4.7
- Firefox
- GLM-5.2
- Kimi K3
- Linux Foundation
- OpenRouter
- Opus 4.8
- State of Open Source AI
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →