paritok-4b-v1
Non-destructive compression gateway for AI coding agents. Cuts token bills 25% on turn 1 to past 85% in long or saturated sessions, and fits ~3× more turns in the same context window. Powered by our open-source code-native 4B model. Drop-in for Claude Code, Cursor, Codex, OpenHands, and any BASE_URL agent.
- stars
- 1,454
- 30-day movement
- +10/day
- Related entries
- 60
- Connections
- 1
A compression gateway that sits between AI coding agents and their LLM endpoints, using an open-source 4B code-native model to compress context non-destructively. It claims 25% token savings immediately and past 85% in long sessions, fitting roughly 3x more turns in the same context window.
You reach for it to cut token costs and extend how much conversation fits in your agent's context window without changing your workflow.
Use it to
- Proxy Claude Code, Cursor, or Codex through the gateway
- Compress long agentic coding sessions to save tokens
- Fit more turns in a saturated context window
- Point any BASE_URL-configurable agent at the gateway
For Developers running AI coding agents who want lower token costs
- Role
- agent-app
- Language
- Python
- Licence
- Apache-2.0
- Forks
- 138
- Open issues
- 5
- Last push
- 2026-09-10