A free, open-source VS Code AI assistant that multiplexes 22+ free-tier LLM providers — auto-routing, auto-failing-over, learning your codebase.
Free LLM tiers (Groq, Cerebras, Google AI Studio, Mistral, NVIDIA NIM, OpenRouter…) are powerful but unreliable on their own — rate limits, downtime, blind spots. TierMux pools them into one self-healing surface. No middleman, no API bill, no babysitting.
What you get:
- Zero setup cost — most providers work without a paid API bill; some need no key at all.
- Auto-failover — rate-limited or down provider? TierMux silently moves to the next. You see no errors, feel no delay.
- Auto-routing — every message is classified and sent to the best model for the job: quick questions, code edits, full agent runs, images.
- Code-aware — silently picks up your style (indentation, quotes, semicolons) and instructs models to match.
- It improves over time — your 👍 / 👎 feedback tunes future routing.
- Install from the VS Code Marketplace (search TierMux) or from source.
- Open the TierMux icon in the Activity Bar → ⚙ Manage Models & Keys → enable providers, add keys, set priority.
- Leave everything on Auto and start typing.
No
settings.jsonediting — everything is point-and-click in the panel.
git clone https://github.com/mainulBUBT/TierMux.git
cd TierMux
npm install
npm run package # produces tiermux-*.vsix
code --install-extension tiermux-*.vsixyour message ──▶ Classify ──▶ Rank ──▶ Send ──▶ Fail-over ──▶ Learn
- Classify — decides what kind of task it is and which model class it needs.
- Rank — orders your enabled providers by priority, skipping rate-limited ones and filtering out models that can't handle the task.
- Send — your message goes directly from VS Code to the provider's API. No TierMux server in between.
- Fail-over — if a provider rate-limits, times out, or refuses, it silently retries the next one.
- Learn — 👍 / 👎 feedback and your edited files tune future routing and style.
| Mode | Behavior |
|---|---|
| Ask | Read-only — answers and explains, never touches your files. |
| Plan | Reads your code, proposes a step-by-step plan, waits for approval. |
| Agent | Full loop — reads files, applies diffs, runs terminal commands, checkpoints for undo. |
Auto mode picks the right mode and model for each message automatically.
22 pre-configured providers, 121+ models out of the box — Groq, Cerebras, Google AI Studio, Mistral, NVIDIA NIM, OpenRouter, Cohere, GitHub Models, SambaNova, SiliconFlow, and more.
Keyless (zero setup): Kilo Gateway · OVH AI Endpoints · Pollinations
Custom endpoints: any OpenAI-compatible URL (vLLM, LiteLLM, Azure OpenAI, Cloudflare AI Gateway…).
- Keys — stored in VS Code's encrypted secret storage. Never written to disk, never synced, never logged.
- No TierMux backend — every request goes VS Code → provider API directly.
- Feedback & stats — local extension storage only, never uploaded.
- Nothing you type is sent to TierMux — only to the provider you (or Auto) selected.
Contributing? Read PUBLISHING.md before pushing.
MIT — free to use, modify, distribute. See LICENSE.