Settings · Custom LLM provider

AI providers / Custom provider

Together AI

Any service with an OpenAI-compatible API: Together, Groq, Fireworks, DeepSeek, a self-hosted vLLM or Ollama. Calls go from our servers, so the address must be reachable from the internet.

Connection

Name

Together AI

API style

Base URL

https://api.together.xyz/v1

API key

sk-••••••••••••3f9aReplace

Stored encrypted. Only admins can see the last 4 characters.

Extra headers

X-Org: … (optional)

Connected. Found 41 models · answered in 380 ms · tool calling works on 12 of them.

Modelspick the ones your agents may use · prices are per 1M tokens and drive the credit counter

	Model	Can do	Input	Output	Context
✓	meta-llama/Llama-4-Maverick	toolsimagesstream	$0.27	$0.85	1M
✓	deepseek-ai/DeepSeek-V3.2	toolsimagesstream	$0.60	$1.70	128k
✓	Qwen/Qwen3-235B-A22B	toolsimagesstream	$0.20	$0.60	256k
	mistralai/Mixtral-8x22B	toolsimagesstream	$1.20	$1.20	64k

We filled prices from the provider where it publishes them — edit any row if your contract differs. Models without tool calling can chat but can’t create posts or read analytics.

Use it foroptional — you can also pick it per agent

Adapting postsTogether · DeepSeek-V3.2

Engage reply draftsTogether · Qwen3-235B

Agents using it switch back to the workspace default