Dev Tools · 2h ago
DeepSeek, Qwen, Kimi, GLM: Which Chinese LLM Saves You Most at Scale?
A CTO's real-world test of four Chinese LLM families reveals huge cost differences: Qwen3-8B and GLM-4-9B cost just $0.01 per million tokens, while Kimi's cheapest model is $3.00. DeepSeek V4 Flash offers strong code generation at $0.25/M. All support OpenAI-compatible APIs, reducing switching costs.
Meridian48 take
The article's vendor-lock-in warning is valid, but the pricing data is from a single aggregator; actual costs may vary with direct API usage.
llm-comparisonchinese-ai-models