Unverified50% confidenceFactExact time
斯坦福的Alpaca以GPT-3.5蒸馏LLaMA，Microsoft的Orca系列以GPT-4蒸馏小模型，验证了高质量轨迹数据对小模型能力迁移的有效性
1
Sources
50%
Confidence
Long-term
Relevance
7/12/2026
First Seen
Sources
Related Claims
VerifiedStanford University once published a research report documenting performance fluctuations of GPT-4 on specific tasks across different time periods.74% similarUnverifiedStanford University's AgentBench benchmark test in 2024 showed that GPT-4-level models have an end-to-end success rate below 30% on complex tasks requiring cross-system coordination.70% similarUnverifiedThe Microsoft Foundry Portal (ai.azure.com) system automatically recommends GPT-4.1 as the starting model.69% similarVerifiedGPT-4和Mixtral等主流大模型也采用了MoE架构68% similarUnverified分层模型策略将 GPT-4o 等高能力模型用于规划决策节点，中间执行步骤使用 GPT-4o-mini 或本地开源模型（如 Llama 3）以控制成本68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/489811API
curl https://kongchang.com/api/v1/knowledge/claims/489811MCP
get_claim(id=489811)