serverless access to meta-llama/Llama-3.3-70B-Instruct-Turbo and deepseek-ai/DeepSeek-V4-Flash-0731: October 22, 2026

Completed

The Together AI API keeps running. meta-llama/Llama-3.3-70B-Instruct-Turbo and deepseek-ai/DeepSeek-V4-Flash-0731 can be called on serverless inference until October 22, 2026, when Together removes them. Together recommends meta-models/Muse-Glimmer-30B or a dedicated endpoint in place of Llama 3.3 70B, and deepseek-ai/DeepSeek-V4.1-Flash in place of DeepSeek-V4-Flash-0731.

Source

Vendor · Shutdown tracker