dsh-our-free-model
Open-source dsh plugin that lets dsh users call hosted DeepSeek, Kimi and other models without a login or API key.
No visits from OpenBeat in the last 30 days yet. Every click on “Visit” is counted here.
Pricing and API checked by an OpenBeat editor on Nov 9, 2026 · facts from the maker’s own site ·
Tool insights
dsh-our-free-model adds a set of hosted models to dsh without an account, sign-up or key. It also opens a local gateway, so other coding tools can use the same models.
Trying hosted models usually means creating accounts, collecting API keys and setting up billing first. This plugin removes that setup for dsh users who want working models quickly.
Best for dsh users who want free hosted models, and for people who want to point Claude Code, Codex or SDK-based scripts at a local gateway. Teams that need stated usage guarantees or strict data handling should look elsewhere.
What it does
dsh-our-free-model is a plugin for dsh, made by Ebony-Vinyl. It gives dsh access to hosted models without a login, sign-up or API key. The models named in the facts are DeepSeek V4.1 Flash, Kimi K3, GLM 5.3 Flash and MiMo V2.6.
The plugin also runs a local gateway. Other tools can send requests to it, and it forwards them to the hosted models. The main use case is running assistants and agents on free models from inside dsh.
Key features
- Installs as a dsh plugin with no account or key.
- The model list is refreshed from upstream each time you refresh it.
- Reasoning effort can be set per model.
- A local port forwards requests in an OpenAI-compatible format.
- The local gateway also offers an Anthropic Messages endpoint.
- Tool output can be compressed to save tokens.
- Named fallback chains let you switch between model lanes.
- A usage dashboard is stored locally on your machine.
Pricing
As of November 2026, the plugin is free. The maker lists a single Free plan at 0, with no login, sign-up or API key required. No usage cap is stated on the maker's page.
That does not mean unlimited use. The upstream free quota is counted per exit IP, and a 429 response pauses requests for a window. The default window is 6 hours.
API and integrations
The plugin has an API in the form of its local gateway. It accepts OpenAI-compatible requests on a forwarding port and has an Anthropic Messages endpoint. The listed integrations are dsh, Claude Code, Codex, the OpenAI SDK, the Anthropic SDK and curl.
The maker does not list its own SDKs. The plugin is open source under the MIT license, and the code is on GitHub. Platforms listed are Web, Windows and macOS. A documentation link and a changelog link are not stated.
Good to know
- Rate limits: a 429 pauses requests for a window, 6 hours by default. Quota is counted per exit IP.
- Privacy: prompts sent through the Kilo channel may be recorded upstream for service improvement.
- macOS: the community build runs on Apple Silicon only and is not Apple-signed.
- LAN relay: it is off by default and needs its own key.
- Package warning: the maker says the unscoped npm package named dsh-our-free-model is third-party. Do not install it.
- Founding date and pricing page: not stated on the maker's site.
FAQ
Is dsh-our-free-model free?
Yes. As of November 2026, the maker lists one Free plan at 0, with no login, sign-up or API key. No usage cap is stated, but upstream free quota is counted per exit IP and a 429 can pause requests.
Do I need an API key to use dsh-our-free-model?
No. The plugin installs into dsh without an account or key. The optional LAN relay is a separate feature, off by default, and it needs its own key.
Which models does dsh-our-free-model offer?
The maker names DeepSeek V4.1 Flash, Kimi K3, GLM 5.3 Flash and MiMo V2.6. The model list is refreshed from upstream each time you refresh it, so it can change.
Can I use dsh-our-free-model with Claude Code or the OpenAI SDK?
Yes, according to the listed integrations. The plugin runs a local gateway with an OpenAI-compatible forwarding port and an Anthropic Messages endpoint. Claude Code, Codex, the OpenAI SDK, the Anthropic SDK and curl are all named.
What happens when I hit the free quota?
Upstream free quota is counted per exit IP. When you hit it, a 429 response pauses requests for a window, which is 6 hours by default.
Releases
No tracked releases yet.
In the news
No coverage yet.