Use a syn: alias to automatically route to the latest recommended model for a given category. These aliases are always up-to-date and are the easiest way to ensure you're using the best available model.

Alias	Resolves To	Category
syn:large:text	hf:deepseek-ai/DeepSeek-V4.1-Flash-1017	Large text model
syn:small:text	hf:zai-org/GLM-4.7-Flash	Small text model
syn:large:vision	hf:moonshotai/Kimi-K3	Large vision model
syn:small:vision	hf:deepseek-ai/DeepSeek-V4.1-Flash-1017	Small vision model

Example:

curl -X POST https://api.synthetic.new/v1/chat/completions \
  -H "Authorization: Bearer ${SYNTHETIC_API_KEY}" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "syn:large:text",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Infinite mode is designed for background agents: it doesn't count towards your rate limits, so you can run tasks in the background as long as you want. It's much slower than regular inference, but it's free: you'll never run out of tokens. Infinite mode is available to subscribers on plans that include Infinite mode.

Alias	Resolves To	Category
infinite:small:vision	hf:deepseek-ai/DeepSeek-V4.1-Flash-1017	Small vision model

Example:

curl -X POST https://api.synthetic.new/v1/chat/completions \
  -H "Authorization: Bearer ${SYNTHETIC_API_KEY}" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "infinite:small:vision",
    "stream": true,
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Infinite mode requires streaming (stream: true), since there may be long pauses in token generation that could cause unstreamed HTTP requests to time out.

These models are included in all subscriptions. Per-token pricing is also available with usage-based billing.

Embedding models convert text into numerical vectors for search, clustering, and other applications.

There's no additional charge for using embeddings, and embeddings requests don't count against your subscription rate limit.

Ready to start using our models? Check out:

• Getting Started Guide - Your first API call
• /chat/completions - Most popular endpoint for conversations

Need help choosing the right model? Join our Discord community for recommendations!