Releases

Together added production rollouts, autoscaling, observability, multi-region deployment, A/B testing, shadow traffic, and a custom-training beta.

Together launched a web-based fine-tuning platform with preference optimization, continued training, deployable checkpoints, and weight ownership.

Together launched self-service, single-tenant dedicated endpoints for hosted and custom fine-tuned models.

Together released a new inference stack with Turbo and Lite endpoints and reported decoding throughput up to four times faster than vLLM.

Together introduced its optimized inference engine across serverless endpoints and dedicated model deployments.

Together launched Together API and Together Compute for inference, fine-tuning, and large-scale training of open models.