Together AI launches Post-Training service at $9 per B200 GPU-hour

Together AI on October 19, 2026 launched Together Post-Training, a service for building custom models via user-defined RL and supervised fine-tuning loops, billed per GPU-minute. Introductory pricing is $9 per B200 GPU-hour for the first six weeks, with Muse Spark Open, GLM 5.3, Kimi K3 and Qwen 3.8-27B available at launch.

• Introductory price: $9 per B200 GPU-hour for the first six weeks
• Supports LoRA and full-weight training with GRPO, CISPO and DPPO losses
• Launch models: Muse Spark Open, GLM 5.3, Kimi K3 and Qwen 3.8-27B
• Includes a Tinker API migration path and multi-LoRA serving after training