Zero-cost access to 6 free large models. No registration, no recharge, 6 free models built-in, multi-key rotation and rate-limit backoff.
Install
# from GitHub (first run asks for allowBuilds approval — follow the hint, retry)
dsh plugin --profile web add github:xiaozhe7772222/dsh-opencode-zen
Any plugin you install runs third-party code with your own permissions — it can read your files, use your credentials, and reach the network, and tool approvals don’t sandbox it. GitHub-sourced plugins also run build scripts at install time — pnpm blocks those until you allow them, so an install can stop with ERR_PNPM_GIT_DEP_PREPARE_NOT_ALLOWED or ERR_PNPM_IGNORED_BUILDS; dsh prints the exact key to add under allowBuilds in your profile’s pnpm-workspace.yaml, and the install works on the next run. Allowing a build is a trust decision: only install sources you trust, and pin a commit (github:owner/repo#sha).
README
Six free LLMs for DeepSeek Harness, zero config, zero cost. Brings the OpenCode Zen free tier into your DSH model picker — no signup, no API key, no billing.
Why?
The conversation you're reading right now is powered by this plugin: DeepSeek V4 Flash on the free tier, for free.
- 💰 Actually free — the official free tier authenticates with the literal key
public; no account, no signup, no API key. - 🧮 Six free models — DeepSeek V4 Flash, Xiaomi MiMo, Tencent Hunyuan, two NVIDIA Nemotrons, and Laguna S 2.1.
- ⚡ Install & go — restart
dsh weband theopencoderoute appears in the model selector; no configuration needed. - 🔑 Stack quotas — pairs with dsh-api-key-pool for round-robin rotation across multiple free accounts, automatically.
- 🛡️ Quota-aware — built-in 429/5xx backoff and request throttling so you never blow through the free quota.
- 🧠 Full parity — streaming, reasoning-content passthrough, and tool calls, same experience as paid models.
Models (6 free models)
| Model | Context window | Notes |
|---|---|---|
deepseek-v4-flash-free |
200k | DeepSeek V4 Flash · reasoning + tool calls, daily driver |
mimo-v2.5-free |
200k | Xiaomi MiMo 2.5 |
hy3-free |
200k | Tencent Hunyuan |
nemotron-3-ultra-free |
131,072 | NVIDIA Nemotron 3 Ultra |
nemotron-3.5-lightning-free |
131,072 | NVIDIA Nemotron 3.5 Lightning |
laguna-s-2.1-free |
200k | Laguna S 2.1 |
Reasoning effort: off / low / high (default) / max.
Installation
dsh plugin --profile web add github:xiaozhe7772222/dsh-opencode-zen
Restart dsh web → Settings → Models → pick provider opencode → choose a free model (start with deepseek-v4-flash-free).
Configuration (optional — zero config by default)
Stack multiple accounts
- Install dsh-api-key-pool.
- Add your keys under the
opencodepool. - The plugin picks them up automatically and rotates round-robin.
Environment variables
Set OPENCODE_ZEN_API_KEY or OPENCODE_GO_API_KEY before starting dsh web.
Nothing configured? It falls back to the official public tier (public).
Troubleshooting
Q: Model returns 429 Too Many Requests? A: The free tier has per-IP rate limits. Wait 30–60 seconds, or install dsh-api-key-pool to rotate across multiple keys automatically.
Q: opencode provider doesn't appear in model selector?
A: Restart dsh web fully (not just refresh). Verify installation with dsh plugin --profile web list.
Q: Which DSH versions are supported?
A: DSH 0.8.0+ with the ctx.llm.registerAdapter API. Older versions may need manual route registration.
Q: Are these models really free forever? A: They use OpenCode Zen's official public free tier. Service availability and quota limits are subject to OpenCode Zen's policies — this plugin is just a client adapter.
How it works
Registers an opencode LLM provider route via ctx.llm.registerAdapter(['opencode'], adapter), exposing the OpenCode Zen free models to session models and sub-agents alike.
License
MIT
Links
More in this category
V1ki/dsh-plugin-subscriptions★ 303
Use ChatGPT (Codex), Claude, and Grok subscriptions as DeepSeek Harness LLM providers, with Settings login, model catalogs, usage, plus image_generate, video_generate, and x_search tools.
Mars-Sea/dsh-commandcode-provider★ 133
Unofficial Command Code LLM provider: registers a `commandcode` route with a live model catalog and reasoning-effort support.
Stormycry-cryp/dsh-AuthInOne★ 105
Adds account login, API and custom Provider setup, model switching, image fallback for text-only models, and token/cost attribution to DeepSeek Harness 47f.
franksong2702/dsh-codex-connect★ 67
Connect ChatGPT OAuth and OpenAI Codex models to DeepSeek Harness, with opt-in search and image tools.
amlyczz/dsh-agy-link★ 30
Google Antigravity (agy CLI) models for DSH — streaming chat with Gemini/Claude/GPT-OSS subscriptions, native tool cards, thinking turns, and in-GUI Google OAuth login.
WSL043/dsh-codex-subscription★ 30
ChatGPT OAuth provider for Codex models with subscription web search, quota and safe reset, image tools, Fast mode, and model-aware context; no API key or Codex CLI.
Community comments
Comments are public GitHub Discussions. Loading them connects to GitHub and Giscus; a GitHub account is required to post.