DeepSeek Reasonix: Native Coding Agent with High Caching and Low Cost
Reasonix is a new terminal-based AI coding agent built natively for DeepSeek models. It's designed around two key constraints: high caching efficiency and low cost — both critical for developers running code generation loops frequently.
Key Details
- Native DeepSeek support — Reasonix is tuned specifically for DeepSeek's architecture, not wrapped around a generic API. This allows optimizations like KV-cache reuse across consecutive requests.
- Aggressive caching — The tool caches intermediate results (prompt embeddings, partial completions) to avoid redundant computation. Early reviews mention “near-instant warm restarts” after the first call.
- Cost — Paired with DeepSeek's newly permanent V4 Pro price discount (from $0.50/M tokens → $0.17/M tokens, per HN thread), Reasonix claims “best-in-class per-request cost” for an agentic coding loop.
- Terminal-native — Runs as a CLI tool (
reasonix --model deepseek-v4-pro). No IDE plugin required. Supports both streaming and batch mode. - Open source — Repository at esengine.github.io/DeepSeek-Reasonix.
📖 Read the full source: HN AI Agents
👀 See Also
Monarch v3: NES-Inspired KV Paging for 78% Faster LLM Inference
Monarch v3 implements NES-inspired memory paging for transformers, achieving 78% faster inference (17.01 to 30.42 tok/sec) on a 1.1B parameter model with nearly zero VRAM overhead. The open-source algorithm splits KV cache into hot and cold regions with compression and promotion mechanisms.
Developer Achieves Sub-Second STT/TTS Latency with Local Whisper and Coqui-TTS Servers
A developer has open-sourced local server implementations for Whisper STT and Coqui TTS that achieve ~0.2s speech-to-text and ~250ms text-to-speech latency, enabling conversational AI agents without cloud dependencies.
Developer builds AI framework with 17 biological principles using Claude Code
A developer created an AI framework called Cognitive Sparks by implementing 17 biological principles like threshold firing and Hebbian plasticity, based on the 1999 book 'Sparks of Genius.' The entire project—22 design docs and 3,300 lines of code—was built in one day using Claude Code, with no human-written code.
Open-Source Web UI for Parallel Claude Code Sessions Using Git Worktree
A developer has built an open-source web UI called CCUI that enables running multiple Claude Code sessions in parallel using git worktree. It runs as a local web server accessible via browser and supports SSH port forwarding for remote development.