Together AI has updated its serverless inference pricing to include latest versions of leading Chinese AI models including MiniMax M2.7, Kimi K2.6, and GLM-5.1 alongside established open-weight options like Gemma 4 and Qwen3.5.
Together AI has refreshed its serverless inference pricing page with updated model listings and new price points. The update includes the latest iterations of high-profile Chinese models: MiniMax M2.7 now priced at $0.30 per million input tokens (with cached hits at $0.06) and $1.20 for output; Kimi K2.6 at $1.20 input and $4.50 output; and GLM-5.1 at $1.40 input and $4.40 output. Batch API pricing is also listed, offering significantly reduced rates for asynchronous workloads. The platform continues to offer a broad range of models spanning chat, vision, image, audio, transcription, embeddings, reranking, and moderation categories. Open-weight models like Gemma 4 31B remain competitively priced at $0.20 input and $0.50 output per million tokens, while the recently listed gpt-oss-120B is priced at $0.15 input and $0.60 output. The LFM2 24B A2B model appears as one of the lowest-cost options on the board at $0.03 input per million tokens. Together AI positions serverless inference as the entry point for most teams, with a migration path to dedicated endpoints as usage scales. The pricing page notes that most teams start serverless before moving to dedicated infrastructure at higher volumes.