Loading the SOTA2 catalog…
Caveman
Caveman builds local context compression, coding-agent tools, and TypeScript and Python middleware. It cuts AI agent token costs through output compression, input compression, model routing, and caching.
- Founded
- Not disclosed
- Headquarters
- Not disclosed
- Company size
- 1–10
- Products tracked
- 6
Products by Caveman
Pricing for Caveman
Per Seat
This is the broad commercial model reported for Caveman. Product-level terms are listed below when available.
- Pricing model
- Free
- Entry option
- Free plan and Free trial
- Starting price
- $0
- Pricing model
- Tiered, Usage Based
- Entry option
- Free plan and Free trial
- Starting price
- $0/month
- Entry option
- Free plan and Free trial
- Pricing model
- Free
- Entry option
- Free plan and Free trial
- Starting price
- $0
- Pricing model
- Free
- Entry option
- Free plan and Free trial
- Starting price
- $0
- Entry option
- Free plan and Free trial
Trust profile
Security and compliance
ISO/IEC 27001
Latest updates from Caveman
Research
Caveman Publishes CaveBench Wrap Benchmark Report
Caveman published a comparative analysis of local AI tooling, including its own system, RTK, Headroom, and context-mode. The report features the CaveBench Wrap benchmark, which assessed input token reduction across 54 runs using Claude Code 2.1.223 and claude-sonnet-5. Caveman reported a 33.2% reduction in input tokens for its system compared to direct Claude Code usage, while maintaining an exact-answer success rate of 18/18. The study emphasizes that tool-side token counting varies by method and provides guidance on selecting compression layers based on specific workflow requirements.
Buyer request
Want to contact this company?
Tell SOTA2 how you would like to connect. We securely save your request for this company's verified team, whether the profile is claimed now or claimed later. Anonymous visitors stay anonymous.