Zed editor setup with Thalam
Install Zed, paste your Thalam API key into the THALAM_API_KEY env var, and ship in under a minute. Every model you list appears in Zed's agent panel — Claude for chat, DeepSeek for tool calls, GPT for refactors — all from one API key.
Setup in three steps
Zed configures LLM providers via settings.json for the model list, and via an environment variable for the API key (Zed never writes keys to the settings file).
Sign up
Create a THALAM account and copy your API key from the dashboard. No credit card required.
Edit Zed settings
Open the command palette (Cmd+Shift+P), run zed: open settings, and paste the snippet below into your settings.json.
Set the API key env var
Export THALAM_API_KEY=tl-live-... in the shell that launches Zed. The provider name (THALAM) maps to <NAME>_API_KEY automatically.
{
"language_models": {
"openai_compatible": {
"THALAM": {
"api_url": "https://api.thalam.ai/v1",
"available_models": [
{
"name": "anthropic/claude-sonnet-4.6",
"display_name": "Claude Sonnet 4.6",
"max_tokens": 200000,
"capabilities": { "tools": true, "images": true }
},
{
"name": "deepseek/deepseek-v3.2",
"display_name": "DeepSeek V3.2",
"max_tokens": 128000,
"capabilities": { "tools": true, "images": false }
}
]
}
}
}
}The provider name THALAM is what you'll see in Zed's model picker, and Zed expects an env var of the same name uppercased + underscored: THALAM_API_KEY. List as many models as you want under available_models — browse ids on the . Set capabilities.images to false for text-only models so Zed routes vision tasks elsewhere.
Verify it works
Open the Zed agent panel (Cmd+Shift+A), pick a THALAM model from the picker, and ask:
Common gotchas
Three issues that account for almost every "it didn't work" report.
"No API key" error
Zed read settings.json before THALAM_API_KEY was exported. Restart Zed from the same shell where you exported the env var, or set the env var system-wide and log out/in. Zed never reads the key from settings.json.
Model not in picker
Check the JSON parses (one trailing comma will silently break the whole settings file). Zed shows a status-bar warning when settings.json is invalid — click it to jump to the offending line.
Wrong key resolution
Zed derives the env var name from the provider name. If you renamed the provider to "Thalam" instead of "THALAM", the env var becomes THALAM_API_KEY either way (uppercased), but spaces become underscores — keep names alphanumeric.
Why use THALAM with Zed
Zed's agent panel can route to any provider that exposes the OpenAI-compatible shape. THALAM puts every leading model behind that single endpoint.
Every leading model
Add as many model entries as you want to available_models — Claude, GPT, Gemini, DeepSeek, Qwen — all under one provider.
Per-model capabilities
Tag each model's capabilities (tools, images) so Zed routes vision and tool-calling work to the right model automatically.
Keys never in settings
Zed reads the API key from env, not settings.json — safe to commit your settings without leaking secrets.
Pay per token
No subscription, no minimums. Same THALAM key works in every other tool you use — Cline, Continue, Cursor, Aider, OpenCode, your own scripts.
Frequently asked questions
What developers ask before wiring Thalam into the Zed editor.
Do I need an API key to use Zed with Thalam?
?
Yes. Zed's OpenAI-compatible provider reads its API key from an environment variable named <PROVIDER>_API_KEY (so THALAM_API_KEY for the THALAM provider). To route Zed through Thalam, create a free Thalam account, copy your API key from the dashboard, and export it as THALAM_API_KEY in the shell that launches Zed. No credit card required to create the key — you only pay when you make requests.
How do I install Zed and connect it to Thalam?
?
Three steps. (1) Install Zed from zed.dev for Mac, Linux or Windows. (2) Sign up at thalam.ai and copy your tl-live-... key. (3) Open the command palette (Cmd+Shift+P / Ctrl+Shift+P), run "zed: open settings", paste the language_models.openai_compatible.THALAM block from the snippet above, then export THALAM_API_KEY=tl-live-... in the shell that launches Zed.
Is Thalam OpenAI-compatible?
?
Yes. Thalam exposes an OpenAI-compatible base URL at https://api.thalam.ai/v1, which is exactly what Zed's openai_compatible provider type expects. Every model Thalam routes to (Claude, Gemini, DeepSeek, Qwen and more) is reachable through the same OpenAI Chat Completions interface and shows up in Zed's agent-panel model picker.
Can I use Thalam with Zed on Mac and Windows?
?
Both, plus Linux. Zed runs natively on macOS, Linux and Windows. settings.json works identically on every platform; only the path differs (~/.config/zed/settings.json on Mac/Linux, %APPDATA%\Zed\settings.json on Windows). The env var THALAM_API_KEY is read the same way on each OS.
Can I switch models without changing my code?
?
Yes. Add as many models as you want to the available_models array under your THALAM provider entry, then pick from Zed's model dropdown in the agent panel. 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.