DeepSeek-Reasonix
DeepSeek-native AI coding agent for your terminal with prefix-cache stability.
Run autonomous coding sessions with cache-aware context maintenance and workspace sandboxing.
DeepSeek-Reasonix is an open-source, DeepSeek-native AI coding agent built around prefix-cache stability. It provides a single local engine accessible via terminal, desktop app, browser, or editor integration.
What it is
It executes autonomous coding tasks using a config-driven architecture specified in `reasonix.toml`. It supports dual-model execution separating executor and planner roles, MCP server tool integration, and per-turn checkpoints.
Who it's for
It suits developers working extensively with DeepSeek or OpenAI-compatible endpoints who want persistent, multi-interface coding agents. It does not suit teams looking for cloud-managed SaaS agent platforms without local setup requirements.
Key features
Config-Driven Setup
Providers, agents, enabled tools, and plugins are all declared in a single reasonix.toml file.
Cache-Aware Maintenance
Startup injects a stable environment summary while stale tool outputs are pruned before summary compaction.
Multi-Model Composable
Run two models concurrently as an executor and planner in separate cache-stable sessions.
Plugin-Driven Architecture
MCP servers contribute tools while Extension Protocol v1 sidecars intercept runtime events and UI.
Up and running in 3 steps
Our honest verdict
What's great
- Single static Go binary distribution with zero runtime dependencies.
- Accessible across terminal, desktop app, web browser, and VS Code via ACP.
- Per-turn checkpoints allow you to read and undo autonomous runs.
- Native support for prefix-cache stability to optimize long sessions.
Where it falls short
- Requires Go 1.26+ or Node 24 with pnpm 10 if you build from source.
- Consumes external API tokens for DeepSeek or other OpenAI-compatible providers.
- Requires initial manual configuration of providers and models.
Pricing & license
Free under its MIT license. The only cost is running it yourself and the API token usage for your chosen model provider.
Open remote AI-training roles
870 open remote AI-training roles, updated November 14, 2026; most pay $20–$170 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
How do I configure my models and providers?
You can run reasonix setup or declare them directly inside the reasonix.toml configuration file.
Can I use editors other than VS Code?
Yes, any editor supporting the Agent Client Protocol (ACP) can interface with the local Reasonix backend.
