Coding Agents

Continue.dev setup with Thalam

Install Continue.dev, paste your Thalam API key into config.yaml, and ship in under a minute. Assign each role its own model — Claude for chat, DeepSeek for autocomplete, GPT for refactors — all from one API key.

Available inVS CodeJetBrains

Install Continue

Setup in three steps

Continue is configured by editing a single YAML file. The file is created automatically the first time you open Continue.

Sign up

Create a THALAM account and copy your API key from the dashboard. No credit card required.

Open config.yaml

Open the Continue panel, click the gear ⚙️ icon, and choose "Open config.yaml" (or edit ~/.continue/config.yaml directly).

Add THALAM models

Paste the snippet below into the models: list, replace tl-live-... with your real key, and save. Continue picks it up live.

~/.continue/config.yaml

models:
  - name: THALAM Claude Sonnet
    provider: openai
    model: anthropic/claude-sonnet-4.6
    apiBase: https://api.thalam.ai/v1
    apiKey: tl-live-...
    roles:
      - chat
      - edit
      - apply

  - name: THALAM DeepSeek (autocomplete)
    provider: openai
    model: deepseek/deepseek-v3.2
    apiBase: https://api.thalam.ai/v1
    apiKey: tl-live-...
    roles:
      - autocomplete

The example lists two THALAM models — Claude for chat / edit / apply, and DeepSeek for autocomplete. Add as many as you want. Browse model ids on the (format is <provider>/<model>).

Verify it works

Open the Continue chat panel, pick "THALAM Claude Sonnet" from the model dropdown, and ask:

Which model are you and which API base URL is serving you?

Common gotchas

Three issues that account for almost every "it didn't work" report.

Model doesn't appear

YAML is whitespace-sensitive. Each model entry under models: must start with - and the indented fields must use two-space indents. If your model is missing, paste the snippet again rather than editing in place.

404 model not found

The model field expects the full id <provider>/<model> — for example anthropic/claude-sonnet-4.6, not claude-sonnet-4.6 alone.

Autocomplete doesn't fire

The model assigned the autocomplete role must support fast, low-latency completions. DeepSeek V3.2, Qwen3, and other open-weight models work well; flagship reasoning models will be too slow.

Why use THALAM with Continue

Continue's strongest feature is per-role model assignment — different models for chat, autocomplete, edit, and embedding. THALAM lets you populate every role from one key.

One key, every model

Claude, GPT, Gemini, DeepSeek, Qwen, Kimi, Llama, Mistral, GLM. List as many as you want in config.yaml — each one routes through the same THALAM endpoint, no extra provider keys.

Per-role optimisation

Flagship for chat, cheap fast model for autocomplete, embedding-tuned model for retrieval. Same config, three roles, three different upstreams.

No lock-in, no minimums

Pay per token. To swap THALAM out, change apiBase + apiKey on each model entry — Continue never knows the difference.

OpenAI-compatible end to end

Streaming, tool calling, structured outputs, vision — every Continue feature works through Thalam's /v1 endpoint without custom adapters.

Frequently asked questions

What developers ask before wiring Thalam into Continue.dev.

Do I need an API key to use Continue.dev with Thalam?

Yes. Continue calls models through whichever provider you list in config.yaml. To route Continue through Thalam, create a free Thalam account, copy your API key from the dashboard, and paste it as the apiKey on each Thalam model entry in your config.yaml. No credit card required to create the key — you only pay when you make requests.

How do I install Continue.dev and connect it to Thalam?

Three steps. (1) Install the Continue extension from the VS Code Marketplace or the JetBrains plugin marketplace. (2) Sign up at thalam.ai and copy your tl-live-... key. (3) Open the Continue panel, click the gear icon, choose "Open config.yaml" (or edit ~/.continue/config.yaml directly), and paste the model snippet above with provider: openai, apiBase: https://api.thalam.ai/v1, your key, and the model id.

Is Thalam OpenAI-compatible?

Yes. Thalam exposes an OpenAI-compatible base URL at https://api.thalam.ai/v1, so Continue's built-in "openai" provider treats it the same way it would treat OpenAI direct — just with a different apiBase and apiKey. 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 Continue.dev on Mac and Windows?

Both, plus Linux. Continue runs wherever VS Code or a JetBrains IDE runs. The config.yaml lives at ~/.continue/config.yaml on Mac/Linux and at C:\Users\<you>\.continue\config.yaml on Windows — the Thalam model entry itself is identical on every platform.

Can I switch models without changing my code?

Yes. List as many Thalam model entries as you want in config.yaml, each with its own role(s) — chat, edit, apply, autocomplete. Switching between Claude, GPT, DeepSeek or any other Thalam-routed model is a config change, not a code change, and Continue picks up edits to config.yaml live without restarting your editor.

Plug in. Pick a model. Ship.

Get an API key in under a minute. No credit card required.