Unverified50% confidenceFactExact time
Meta's LLaMA series, evolving from LLaMA to LLaMA 3.1 405B, opened the door to open-source large models.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/5/2026
First Seen
Valid until: 11/3/2026
Sources
高性能低成本AI：技术普惠时代的关键拐点
Related Claims
UnverifiedMeta released Muse Spark 1.1, the first model in the Spark series to offer an API interface.59% similarUnverifiedMeta承诺其旗舰模型Muse Spark 1.2的全部权重将在几周内开源58% similarUnverifiedMeta 的 Llama 3 系列和 Mistral 的 Mixtral 均采用 MoE 类似设计57% similarUnverifiedTesseract introduced an LSTM-based recognition engine starting from version 4.0, and version 5.x supports Marathi (the mar language pack).51% similarUnverifiedWindsurf launched its own self-developed programming large model earlier than Cursor, and the SWE series has iterated from version 1.0 to 1.5.51% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/687491API
curl https://kongchang.com/api/v1/knowledge/claims/687491MCP
get_claim(id=687491)