AI Providers

Configure the AI models that power your provisioning interviews.

Supported Providers

ProviderAPI EndpointKey Setup
OpenCode Goopencode.ai/zen/go/v1OpenCode subscription
OpenCode Zenopencode.ai/zen/v1OpenCode subscription
DeepSeekapi.deepseek.com/v1DeepSeek Platform

How Model Lists Work

Ignite caches available models in a local SQLite database on startup. Models refresh automatically every 15 minutes in the background. If you haven't configured an API key yet, Ignite shows a default model list so you can still browse and plan before setting up your provider.

Available Models by Provider

OpenCode Go

Without an API key, the following models appear by default. With a key configured, the live model list replaces these on the next sync.

GLM-5.2Zhipu AI's flagship model
GLM-5.1Zhipu AI's prior generation
Kimi K2.7 CodeMoonshot AI's code-focused model
Kimi K2.6Moonshot AI's general purpose model
MiMo-V2.5Minimax's balanced multimodal
MiMo-V2.5-ProMinimax's most capable variant
MiniMax M3MiniMax's latest generation
MiniMax M2.7MiniMax's previous generation
Qwen3.7 MaxAlibaba's frontier model
Qwen3.7 PlusAlibaba's balanced variant
Qwen3.6 PlusAlibaba's prior generation
DeepSeek V4 ProDeepSeek's frontier tier (49B active)
DeepSeek V4 FlashDeepSeek's fast tier

OpenCode Zen

Curated selection of GPT-5, Claude, Gemini, Grok, and DeepSeek models. Without an API key, the following models appear by default. With a key configured, the live model list replaces these on the next sync.

GPT-5.5OpenAI's latest flagship
GPT-5.5 ProOpenAI's most powerful
GPT-5.4GPT-5.4 generation
GPT-5.4 ProGPT-5.4 enhanced
GPT-5.4 MiniGPT-5.4 compact
GPT-5.4 NanoGPT-5.4 smallest
GPT-5.3 CodexCode-focused generation
GPT-5.2Prior generation
GPT-5.1Stable release
GPT-5.1 CodexCodex generation
GPT-5Legacy GPT-5
GPT-5 CodexLegacy codex
GPT-5 NanoFastest GPT-5
Claude Fable 5Anthropic frontier
Claude Opus 4.8Latest Opus
Claude Opus 4.5Stable Opus
Claude Sonnet 4.5Latest Sonnet
Claude Sonnet 4Stable Sonnet
Claude Haiku 4.5Fast Claude
Gemini 3.5 FlashLatest Gemini
Gemini 3.1 ProStable Gemini
Gemini 3 FlashFast Gemini
Grok Build 0.1xAI's model
DeepSeek V4 ProDeepSeek frontier
DeepSeek V4 FlashDeepSeek fast
GLM-5.2Zhipu AI
GLM-5.1Zhipu prior gen
GLM-5Zhipu legacy

DeepSeek

DeepSeek FlashFast inference, 1M context, $0.14/$0.28 per 1M input/output
DeepSeek ProFrontier reasoning, 1M context, $0.435/$0.87 per 1M (75% permanent discount)

API Key Security

All API keys are stored in your OS keychain (macOS Keychain), never in plaintext configuration files. Ignite reads keys from the keychain at runtime. The ~/.ignite/config.json file stores only provider metadata and preferences — no secrets.