OpenCode setup with Thalam
Install OpenCode (sst), drop your Thalam API key into opencode.json, and ship in under a minute. Claude, GPT, Gemini, DeepSeek, Qwen and more — accessible through the same OpenCode TUI you already use.
Setup in three steps
OpenCode reads opencode.json from the project root or from ~/.config/opencode/opencode.json globally.
Sign up
Create a THALAM account and copy your API key from the dashboard. No credit card required.
Add the provider config
Drop the JSON snippet below into opencode.json (project root) or ~/.config/opencode/opencode.json (global).
Connect & pick a model
In the OpenCode TUI run /connect, enter "thalam" as the provider id, paste your API key, then run /models to pick a model.
{
"$schema": "https://opencode.ai/config.json",
"provider": {
"thalam": {
"npm": "@ai-sdk/openai-compatible",
"name": "THALAM",
"options": {
"baseURL": "https://api.thalam.ai/v1"
},
"models": {
"anthropic/claude-sonnet-4.6": {
"name": "Claude Sonnet 4.6"
},
"deepseek/deepseek-v3.2": {
"name": "DeepSeek V3.2"
}
}
}
}
}Add as many model entries as you want under models. Each key is the THALAM model id (<provider>/<model>) — browse them on the . The name field is just the friendly label OpenCode shows in the picker.
Verify it works
Inside the OpenCode TUI, ask the active model:
Common gotchas
Three issues that account for almost every "it didn't work" report.
Provider not found
The npm field must be exactly @ai-sdk/openai-compatible. OpenCode pulls the package on first use; if it can't reach npm the provider won't load — check your network and run OpenCode again.
Auth not picked up
OpenCode stores keys in ~/.local/share/opencode/auth.json. If /connect never ran, no key is there. Either run /connect in the TUI or set "apiKey": "{env:THALAM_API_KEY}" in your provider's options and export the env var.
Model id missing in /models
You added the model under the wrong provider key, or the JSON has a trailing-comma syntax error. Check models sits under provider.thalam and the JSON validates.
Why use THALAM with OpenCode
OpenCode runs as a TUI agent — model choice and per-call cost dominate any meaningful session.
Every leading model, one provider
List as many model ids as you want under the same THALAM provider entry. Switch via /models in the TUI without restarting.
Project vs global config
opencode.json in the project root overrides ~/.config/opencode/opencode.json — different repos can default to different models with the same THALAM key.
Pay per token
No subscription, no minimums. Auth lives in ~/.local/share/opencode/auth.json (gitignored by OpenCode), so it doesn't leak into commits.
OpenAI-compatible end to end
Streaming, tool calling, structured outputs, vision — every OpenCode feature works through Thalam without a custom adapter.
Frequently asked questions
What developers ask before wiring Thalam into OpenCode (sst).
Do I need an API key to use OpenCode with Thalam?
?
Yes. OpenCode reads provider credentials from ~/.local/share/opencode/auth.json (written by /connect inside the TUI) or from an env var referenced in opencode.json. To route OpenCode through Thalam, create a free Thalam account, copy your API key from the dashboard, then run /connect in the TUI and paste it. No credit card required to create the key — you only pay when you make requests.
How do I install OpenCode and connect it to Thalam?
?
Three steps. (1) Install OpenCode (sst) — see opencode.ai/docs/ for the platform installer. (2) Sign up at thalam.ai and copy your tl-live-... key. (3) Drop the provider block from the snippet above into opencode.json (project root) or ~/.config/opencode/opencode.json (global), then inside the TUI run /connect, enter "thalam" as the provider id, paste your key, and pick a model with /models.
Is Thalam OpenAI-compatible?
?
Yes. Thalam exposes an OpenAI-compatible base URL at https://api.thalam.ai/v1, and OpenCode connects to it via the @ai-sdk/openai-compatible npm package — exactly the integration shape OpenCode uses for every OpenAI-compatible provider. Every model Thalam routes to (Claude, Gemini, DeepSeek, Qwen and more) is reachable through the same OpenAI Chat Completions interface.
Can I use Thalam with OpenCode on Mac and Windows?
?
Both, plus Linux. OpenCode runs in any modern terminal — macOS, Windows (PowerShell or WSL), Linux. opencode.json works the same on every platform; only the global config path differs (~/.config/opencode/opencode.json on Mac/Linux vs. %USERPROFILE%\.config\opencode\opencode.json on Windows).
Can I switch models without changing my code?
?
Yes. List as many model ids as you want under the same Thalam provider entry in opencode.json, then run /models in the TUI to switch between them at runtime. Switching between Claude, GPT, DeepSeek or any other Thalam-routed model is a config change, not a code change.
Plug in. Pick a model. Ship.
Get an API key in under a minute. No credit card required.