Unverified90% confidenceFactTime unknown
通义零码的MoE架构激活比例约为7.3%（350亿/4800亿），推理成本接近一个350亿参数的密集模型
1
Sources
90%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Qwen Code CLI实测：阿里开源AI编码代理安装配置全指南
bilibiliAI-seeker
Related Entities
Related Claims
Unverified在标准云服务器上，一个3.8B参数模型的单次推理成本约为70B模型的1/2071% similarUnverifiedFine-tuning a 7-billion-parameter model can be done on a consumer-grade GPU, potentially at a cost of only a few hundred dollars69% similarUnverifiedThe inference cost of flagship models ranges from $3 to $15 per million tokens, far exceeding the marginal cost of traditional SaaS services.68% similarUnverifiedTencent open-sourced the Hunyuan (HY3) model with a parameter scale of approximately 299 billion.68% similarUnverifiedMainstream model inference costs are priced per million tokens: GPT-4o is approximately $5–15, Claude 3.5 Sonnet is approximately $3–15, and the total token count for multi-step agent tasks is often 10–50 times that of a single conversation.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/23567API
curl https://kongchang.com/api/v1/knowledge/claims/23567MCP
get_claim(id=23567)