Meta's model card lists 'Model Release Date: 70B Instruct: December 6, 2024'; TechCrunch's[5] report is dated December 6, 2024 and OpenRouter[6] lists meta-llama/llama-3.3-70b-instruct as created the same day.
Meta releases Llama 3.3 70B, a text-only 70B instruction-tuned model that Meta says delivers the performance of the 405B Llama 3.1 at lower cost, with a 128K context window and open weights; Meta's chart prices it at $0.1 per 1 million input tokens.
Jiyue