what is latest Chinese models that lowers price and improve performance?
AI Models & CapabilitiesAI Geopolitics
Recent Chinese AI models from March 2026 exemplify efforts to lower prices while enhancing performance, challenging Western dominance. Z.ai's GLM-5 Turbo, a proprietary variant of the open-source GLM-5, offers faster processing for agent-driven tasks like automation and tooling at reduced costs compared to its base model and U.S. competitors [7]. On the same day, March 19, MiniMax released the proprietary M2.7, a self-evolving model building on the earlier M2.5, which already matched leading benchmarks in coding at $1.20–$2.40 per million output tokens—far below Opus 4.6's $25—while handling significant internal productivity tasks [8][9]. Xiaomi's MiMo-V2-Pro, a 1-trillion-parameter model, also launched on March 19, achieving benchmarks near GPT-5.2 and Opus 4.6 at about one-sixth to one-seventh the API access cost of U.S. equivalents, making advanced AI more accessible [12].
Sources
- Chinese AI Models Challenge Western Pricing — Daily Brew
- Chinese AI Models Drive Down Prices in Global Competition — The AI Daily Brief
- Z.ai Launches GLM-5 with Competitive Pricing — The Rundown AI
- Chinese AI Model Challenges US Margins — GAI Insights Newsletter
- Impressive benchmarks for the new Chinese LLM. The system card notes some gaps with US closed source models in code generation & wide knowledge, so be interested to see it in operation. — @emollick
- Z.ai's GLM-5 Challenges Western AI with Competitive Pricing — The Rundown AI
- Z.ai Debuts Faster, Cheaper GLM-5 Turbo Model — VentureBeat
- MiniMax's M2.5 Offers Low-Cost Frontier AI Coding — The Rundown
- New MiniMax M2.7 Proprietary AI Model — VentureBeat
- ByteDance Undercuts AI Model Pricing — The Rundown AI
- China vs. US in the AI Race: How China Is Closing the Gap - Medium
- Xiaomi Stuns with New MiMo-V2-Pro LLM — VentureBeat
Related questions
- →What is retrieval-augmented generation (RAG), and why is it important for enterprise AI deployment?
- →How should non-technical executives evaluate and compare AI model performance benchmarks?
- →What is multimodal AI, and why does it matter for practical business applications?
- →How quickly are AI capabilities improving, and is there credible evidence that the pace of progress is slowing?