Together AI deprecates Qwen3 models and launches Dedicated Container Inference (DCI)

AI Impact Summary

Together AI is deprecating several models, including Qwen3-VL-8B-Instruct and Qwen3-235B-A22B-Thinking, signaling a shift in their model portfolio. The transition to prepaid billing and dynamic rate limits introduces a new pricing structure, impacting cost management for users. Furthermore, the launch of Dedicated Container Inference (DCI) provides a new deployment option for custom models, representing a significant change in Together AI’s infrastructure offerings.