Semantic Caching with GPTCache for Offline Mode

[mikejmorgan-ai]

Problem

Every query hits LLM API, costing money and requiring internet. GPTCache enables 70% cost reduction.

Solution

• Cache responses with semantic similarity matching
• cortex cache stats - show hit rate, savings
• cortex --offline flag for cached-only mode
• SQLite backend, LRU eviction

Bounty: $65 (+ $65 bonus after funding)

Paid on merge to main.

Skills: Python, GPTCache, SQLite, Embeddings

[Sahilbhatane]

attempting