Made with these models
The 4 models this integration runs.
deepseek-ai/DeepSeek-V4-Pro on together
deepseek-ai/DeepSeek-V4-Protext-to-text
meta-llama/Llama-3.3-70B-Instruct-Turbo on together
meta-llama/Llama-3.3-70B-Instruct-Turbotext-to-text
Qwen/Qwen3.5-9B on together
Qwen/Qwen3.5-9Btext-to-text
Qwen/Qwen3.7-Plus on together
Qwen/Qwen3.7-Plustext-to-text
deepseek-ai/DeepSeek-V4-Pro on together
deepseek-ai/DeepSeek-V4-Protext-to-text
meta-llama/Llama-3.3-70B-Instruct-Turbo on together
meta-llama/Llama-3.3-70B-Instruct-Turbotext-to-text
Qwen/Qwen3.5-9B on together
Qwen/Qwen3.5-9Btext-to-text
Qwen/Qwen3.7-Plus on together
Qwen/Qwen3.7-Plustext-to-text
Open-weight models — Llama, Qwen, DeepSeek — on serverless inference.
- Open-weight models without running your own GPU
- Llama 3.3 70B, Qwen 3.5/3.7, DeepSeek V4 Pro
- Symmetric input/output pricing on several models
- Charged on Together-reported token usage
- Bring your own key, or run on the platform key
Together hosts open-weight models — Llama, Qwen, DeepSeek and others — on serverless inference, which makes it the practical way to use an open model without operating a GPU.
It is the natural counterpart to the closed frontier models on this marketplace: same LLM Worker node, same graph, but weights you could in principle self-host later. Several models price input and output symmetrically, which makes cost easy to reason about for generation-heavy work.
Charged on the tokens Together reports, per model.
llmopen-weightsllamaqwendeepseektogetherserverless