OpenCode Bridge 0.4.2 — no more surprise LM Studio launches

Fix: opening Bridge launched LM Studio

When LM Studio's server was off, Bridge listed its models with the lms command, which silently boots LM Studio in the background. Bridge now reads your downloaded LM Studio models straight from disk. LM Studio starts only when you launch Claude or Codex with an LM Studio model.

Offline LM Studio models are listed with a hint. Turn on LM Studio's server and they become usable automatically.

SHA-256 6dbfee5e79373d020d69428666b2de8817fa9a7f3bc6cc69ced0e44c534a6fd4

OpenCode Bridge 0.4.1 — models can edit files

Fix: models couldn't edit files

Free and other OpenCode-agent models (for example DeepSeek V4.1 Flash Free in Codex) had every edit declined, because the agent's tool policy defaulted to Read only.

• The default is now Edit & run: edits and commands are allowed inside your project folder. Anything outside it is still declined.
• If something is blocked, the model now says exactly where to change it (Settings → OpenCode agent), and the tool note reads blocked: ….
• A new Allow edits button on the Launchpad switches the policy in one click.

Verified live: DeepSeek V4.1 Flash Free fixed a bug in calc.py through Bridge.

Already running 0.4.0? Open Settings → OpenCode agent and choose Edit & run, or click Allow edits after updating.

SHA-256 9be05693782fa427c4fb573293549184af6228b8ce68d9c5084af16ecfc7757a

OpenCode Bridge 0.4.0 — Codex names, your models

Codex names, your models

Just like Claude Desktop, Codex now keeps its own model names: GPT-6-Luna, GPT-5.6-Terra, GPT-5.6-Luna and so on. The backend you choose answers each one, with its own effort level.

• Bridge reads the names from your Codex install. Pick a model and effort for each, and mark one as the default.
• Switch models in Codex as usual (/model, -m gpt-5.6-terra, or the app's picker), and the matching backend answers.
• Verified live: gpt-6-luna → DeepSeek V4.1 Flash Free, gpt-5.6-terra → Nemotron 3 Super (free).

Also new

• 🔔 An update notice in the sidebar when a newer release is out.
• 📋 A copy button for codex --profile opencode-bridge.
• ⌨️ ⌘1 ⌘2 ⌘3 to switch pages.

Install

Download OpenCode-Bridge-0.4.0-arm64.dmg (Apple Silicon), drag it to Applications, then right-click → Open the first time (the build is unsigned).

SHA-256 b68164d1a1dcfc3bf2a8a374a9866880e68192401a4e152f158211e642d43b10

🌐 https://opencode-bridge-gilt.vercel.app

OpenCode Bridge 0.3.0

Any model. Your apps. Zero friction.

Run free OpenCode Zen models, Ollama, LM Studio, OpenRouter, NVIDIA and every OpenCode provider inside Claude Desktop and Codex.

Highlights

• 🆓 Free OpenCode Zen models (Big Pickle, DeepSeek V4.1 Flash Free, GLM-5.3 Flash Free, Nemotron…) run through a private OpenCode agent, which is the supported way.
• 🎭 Claude names, your models. Map Sonnet 5, Opus 5, Haiku 4.5, Fable 5 and Sonnet 4.6 to any backend.
• 🧩 Codex CLI profile and Codex app support, with exact undo.
• 🎚️ Per-model effort from Low to Max, snapped to levels each model supports.
• 🔄 Live library. New Ollama, LM Studio and OpenCode models appear within seconds.
• 🎨 Premium light and dark redesign with real provider logos.

Install

1. Download OpenCode-Bridge-0.3.0-arm64.dmg below (Apple Silicon).
2. Drag OpenCode Bridge to Applications.
3. The build is unsigned: the first time, right-click the app → Open.

Requires the OpenCode CLI. Claude Desktop, Codex, Ollama and LM Studio are optional.

SHA-256 2d3af3bfe6d95b59d3e46214a5015165fe5671e3bf12b5bd7092b9bf995c8638

🌐 https://satiricalguru.github.io/Opencode-bridge/