Vinci in OpenCode
OpenCode is a popular open-source terminal AI coding agent. It supports any OpenAI-compatible provider, so adding Vinci is just a config block — and because requests route through the Vinci gateway, every call gets the Vinci character layer, free-tier metering, and ZDR.
1. Get a key
In the Vinci app: Account → API keys → Issue a vinci_live_… key, then expose it:
export VINCI_API_KEY=vinci_live_...2. Add Vinci as a provider
Add this to opencode.json in your project, or to the global config at
~/.config/opencode/opencode.json:
{
"$schema": "https://opencode.ai/config.json",
"provider": {
"vinci": {
"npm": "@ai-sdk/openai-compatible",
"name": "Vinci",
"options": {
"baseURL": "https://vinci.getsimpledirect.com/api/v1",
"apiKey": "{env:VINCI_API_KEY}"
},
"models": {
"vinci-piccolo": { "name": "Vinci Piccolo (4B)", "limit": { "context": 131072, "output": 8192 } },
"vinci-bozza": { "name": "Vinci Bozza (9B)", "limit": { "context": 131072, "output": 16384 } },
"vinci-tela": { "name": "Vinci Tela (27B)", "limit": { "context": 131072, "output": 32000 } }
}
}
}
}npm— use@ai-sdk/openai-compatible(OpenCode is built on the AI SDK).options.baseURL— the Vinci gateway.options.apiKey—{env:VINCI_API_KEY}reads your exported key (or paste the key directly).models— the Vinci model ids you want to expose.
3. Run
opencodeThen run /models and pick a Vinci model (e.g. Vinci Piccolo). That’s it.
Local / self-hosted
To point OpenCode at a local Vinci app (or your own deployment), change the base URL:
"options": { "baseURL": "http://localhost:3000/api/v1", "apiKey": "{env:VINCI_API_KEY}" }OpenCode’s config schema evolves — if a key name changes, check opencode.ai/docs . The integration itself is just “an OpenAI-compatible provider pointed at the Vinci gateway,” which stays stable.