Cost Optim. — Free AI Tools

Save 80%+ with caching, routing, observability.

• GPTCache — Open-source semantic LLM cache — 61–69% hit rate, 97%+ accuracy, saves $2K+/day
• Redis Semantic Cache — Enterprise vector semantic caching with context-enabled search
• Martian — Mechanistic interpretability routing — claims up to 98% cost savings; $9M backed
• Unify.ai — Quality/cost/latency sliders — live benchmarks updated every 10 minutes
• RouteLLM — Open-source strong/weak model router from UC Berkeley LMSYS
• Helicone — One-line proxy — cost tracking, 95% cache savings, 100+ model providers
• Langfuse — Framework-agnostic LLM tracing — generous free tier, European privacy
• LiteLLM — Unified OpenAI-format interface for 100+ providers — proxy server
• PricePerToken.com — Real-time LLM pricing across all providers — MCP server for Claude Code
• Arize Phoenix — Open-source LLM observability — traces, evaluations, experiments, prompt tracking in a single dashboard
• Not Diamond — AI model router — picks the cheapest model that meets your quality bar
• Braintrust — LLM evaluation + optimization platform — open-source, self-hostable