Intel CEO calls Chinese AI models 'good enough' and cheaper
Intel CEO Lip-Bu Tan named DeepSeek, Moonshot's Kimi and Z.ai's GLM at a DBS panel, saying their performance is "already good enough," their token costs are lower and adoption has been "very fast," while adding that the market should decide which model fits which use case. Third-party price trackers put DeepSeek V4.1 Flash at $0.15 per million input tokens and $0.60 per million output tokens in off-peak hours, versus $2 and $10 for OpenAI's GPT-6.1 Sol, though DeepSeek doubles its rates at peak times, Kimi K3's output price is also $10, and GLM 5.3's $4.40 output price is about 44% of GPT-6.1 Sol's. Consulting firm Digital Applied, citing OpenRouter rankings, said Chinese AI providers together passed 45% of weekly token usage in early April, up from under 2% a year earlier, but attributed the rise to per-token cost and specific strengths such as coding and long context rather than benchmark leadership.
If 'good enough and cheap' becomes the default buying standard, enterprises may route workloads by difficulty, reshaping competition between frontier models and low-cost tiers.