- Category
- AI
- Website
- github.com
- Language
- English
- Listed
- Jul 16, 2026
- Stars
- 112
- Forks
- 10
swellweb/reame ↗, C++, MIT, Last commit 25d ago
Reame is a CPU inference server that leans on disk instead of a GPU. Prompts, prefixes and past generations are cached as they are produced, so requests that share a beginning get cheap fast: request #100 costs a fraction of request #1. The API is OpenAI-compatible and llama.cpp sits underneath, which keeps the deployment target modest. Free tiers, shared VPS instances and 2-core ARM boxes are all listed as places it runs.