Our Free Model
A dsh plugin that gives access to hosted AI models without a login or API key.Ebony-Vinyl/dsh-our-free-model
- Stars
- 56.6k
- Forks
- 1.5k
- Licence
- MIT
- Language
- JavaScript
- Last update
- 4 days ago
- Type
- Dev tool
What it is
This is a plugin for DeepSeek Harness (dsh), an AI agent host. Once installed, it adds free hosted models such as DeepSeek V4.1 Flash and Kimi K3 to dsh's model picker, with no sign-up or API key required. It also includes a local OpenAI-compatible port so other tools on the same machine can call those models. The README says the free models come from the OpenCode Zen and Kilo AI public gateways, and that the plugin is at beta status.
- Pick a free hosted model from the dsh model selector and use it for coding or chat inside dsh
- Point a local tool such as an OpenAI SDK client at the plugin's local base URL and API key
- Chain several models into a fallback route so a call moves to the next model if one fails before producing output
- Track token usage and response speed per model in the plugin's local dashboard
Great for
- Developers who already use dsh and want to try models without creating accounts or keys
- People who want to compare several hosted models inside one agent tool
- Users who want a local OpenAI-compatible endpoint for other tools on their own machine
- Testers who are comfortable with a beta plugin and can tolerate model availability changing
Not ideal for
- Anyone sending sensitive or confidential code or text: the README says the Kilo free pool may record prompts to improve the service
- People who need stable, guaranteed uptime, since the project is marked beta and model lists refresh from upstream
- Users who do not run dsh, since the plugin only works inside that host
- People who want to install from npm: the README warns that the npm package named dsh-our-free-model is third-party and differs from this repository
How to get it
- 1
See the README
You'll need: DeepSeek Harness (dsh) host; the README badge lists kernel versions 0.1.5 to 0.1.7-rc.2
Install the plugin into a dsh web profile from a local source directory
dsh plugin --profile web add /绝对路径/dsh-our-free-model - 2
Run it
- Run the install command above, replacing the path with the absolute path to the repository and the profile name with yours
- Restart the dsh application once
- Open the model picker and look for models in the Our Free Model group
- Open Settings, then Our Free Model, to check which models are currently available from your network
- 3
Get the most out of it
- Use the Settings page model list to see per-model availability, vision support, context window, and thinking-level output limits
- Build a named route (combo) that chains several models for automatic fallback when one hits a quota
- Enable the tool-output compression feature (RTK) in the Settings page to reduce token use on large git diffs or logs, as the README describes
- Use the local OpenAI-compatible endpoint with the API key from the Settings page to connect other tools on the same machine
Downloads
Strengths
- No login, sign-up, or API key is needed to start, according to the README
- Ships as a plugin without changing the dsh kernel and with no build step, per the README
- Includes a local dashboard and a local OpenAI-compatible endpoint, and the README says usage data stays on the machine
Trade-offs
- The 'free with no usage cap' claim comes from the project itself and depends on upstream gateways, which can change or restrict access
- Some features, including the EAC channel, require a GitHub login and a star on the repository before conversations are allowed, according to the README
- Prompts sent through the Kilo free pool may be recorded by the upstream provider, so the README advises against sending sensitive content
How it compares
| Project | Stars | Type | Level | Updated | Best for |
|---|---|---|---|---|---|
| dsh-our-free-model (this) | 56.6k | Dev tool | beginner | 4 days ago | Developers who already use dsh and want to try models without creating accounts or keys |
| ollama | 186k | App | beginner | 4 days ago | Developers who want a local model API for their own apps |
| dsh-desktop | 34.5k | App | beginner | 4 days ago | People who want to run DeepSeek Harness without using the command line |
| deepseek-harness | 269k | Dev tool | intermediate | 7 days ago | Developers who want to experiment with an agent harness they can extend with plugins |
| learn-harness-engineering | 22.9k | Learning resource | beginner | 14 days ago | Developers new to building or controlling AI coding agents |
| dsh-desktop | 17.1k | App | beginner | 4 days ago | People who want to run the DeepSeek Harness agent as a regular desktop app on macOS or Windows x64 |
Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.