How does Together AI compare with the alternatives?
Both list gpt-oss-120b at US$0.15 input and US$0.60 output per 1M tokens. Pick Together AI for the lower listed on-demand H100 rate for a dedicated deployment. Pick Fireworks AI when prompts must not be stored by default, since it states Zero Data Retention by default.
By GILD's calculation, at 1,000,000 gpt-oss-120b requests a month of 2,000 input and 500 output tokens, Together costs US$600, and OpenRouter pinned to Together's endpoint adds a US$33 credit fee. Pick Together AI to hold the host and buy dedicated GPUs. Pick OpenRouter to route the same model across hosts.
Both list gpt-oss-120b at US$0.15 input and US$0.60 output per 1M tokens and a 50% batch discount on the models their batch pages list. Groq does not retain inference data by default, while Together stores prompts until an admin turns storage off.
