An OpenAI-compatible coding model optimized for the idea → app stage. It plans, decomposes, scaffolds, and implements cohesive greenfield software—without drifting into generic chat or over-indexing on late-cycle refactors.
Provide a product objective. The model derives architecture, data surfaces, auth boundaries, and build milestones—then executes them in a governed sequence.
Multi-stage planner emits a dependency graph, updates strategy as constraints shift, and preserves intent across iterations.
Optimizes for clarity, modular cohesion, guardrails, and extension—output is meant to be kept, not rewritten later.
Use the familiar OpenAI Chat Completions schema—just point your SDK or tooling at the TerminalHQ base URL. The model takes over planning, scaffolding, dependency shaping, and incremental file synthesis.
// OpenAI-compatible (just change baseURL)
import OpenAI from 'openai'
const client = new OpenAI({
apiKey: process.env.TERMINAL_API_KEY,
baseURL: 'https://api.terminalhq.com/v1',
})
async function main() {
const stream = await client.chat.completions.create({
model: 'vincent-1',
messages: [
{
role: 'user',
content:
'Build me an AI chatbot where i can upload files and it can answer questions about them.',
},
],
stream: true,
})
}
Upcoming milestones include multi-repo refactors, autonomous test synthesis, telemetry-informed iteration loops, and secure schema + migration orchestration. V1 is the substrate for a continuously adapting engineering system—not a conversational toy.