Guides
Connect your tools to Synap
The same 3 steps every time: get access, point the tool at Synap, run it. Five of the coding tools also have a short video of the real thing, and every snippet on every page was run against the live API before it was published.
Coding tools
Video, 1:03
Claude Code
One setup block in your shell, then run claude.
Video, 1:03
Codex CLI
A Synap provider block in ~/.codex/config.toml.
Video, 1:05
Cline
One command: base URL, key and model.
Video, 1:04
Kilo Code
A custom OpenAI-compatible provider in kilo.json.
Video, 1:04
Continue
One model entry in config.yaml.
opencode
One provider block in opencode.json.
aider
Two env vars. Pin Python 3.12.
goose
Four env vars, no config file at all.
SDKs and frameworks
OpenAI SDK
Python and Node, unchanged: set the base URL and key.
LangChain
ChatOpenAI with a different base URL.
Vercel AI SDK
One provider object: base URL, key, done.
LiteLLM
The openai/ prefix, as the SDK or behind the proxy.
LlamaIndex
OpenAILike, then Settings.llm for the whole app.
No-code and automation
n8n
The built-in OpenAI credential, with a Base URL.
Flowise
ChatOpenAI Custom: Base Path and Model Name.
How-to
Switch from OpenAI
Change the base URL and the model id. Nothing else moves.
Switch from OpenRouter
Which ids change, and the routing fields we ignore.
Choosing a model
Price, context and capability, from the live catalogue.
Tool calling and streaming
The full round trip, and the last chunk that trips loops.
Images with Qwen3 VL
Send an image as base64, or as a URL we fetch.