Together AI
Access hundreds of open-source models through Together's inference API.
Configuration
| Field | Required | Example |
|---|---|---|
| Application Name | Yes | my-together-app |
| Model Name | Yes | meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo |
| API Key | Yes | Your Together API key |
Getting Your API Key
- Go to Together AI
- Sign up or log in
- Navigate to Settings → API Keys
- Copy your API key
Popular Models
| Model | Description |
|---|---|
meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8 | Llama 4 flagship MoE model (400B params, 17B active) |
meta-llama/Llama-4-Scout-17B-16E-Instruct | Llama 4 multimodal with 10M context |
deepseek-ai/DeepSeek-V3.1 | Hybrid thinking/non-thinking mode (671B params) |
deepseek-ai/DeepSeek-R1 | Full reasoning model (87.5% AIME) |
deepseek-ai/DeepSeek-R1-Distill-Llama-70B | Free distilled reasoning model |
Qwen/Qwen3-235B-A22B-Instruct-2507 | Latest Qwen3 with 1M token context |
meta-llama/Llama-3.3-70B-Instruct-Turbo | High-quality text model |
For the complete list of 200+ available models, visit the Together AI Models and API Reference.
Advanced Parameters
| Parameter | Type | Description |
|---|---|---|
max_tokens | integer | Maximum tokens in response |
temperature | float | Randomness |
top_p | float | Nucleus sampling |
top_k | integer | Top-k sampling |
repetition_penalty | float | Penalize repetition |
stop | array | Stop sequences |
Setup Steps
- Navigate to AI Applications → New Application
- Select Model Providers tab
- Click Together
- Enter your Application Name and API Key
- Select your desired Model Name
- Configure Advanced Settings (optional)
- Click Test Response to verify (optional)
- Review and submit