Groq vs Not Human Search

Groq runs open models such as Llama, Mixtral, and Gemma on its own LPU inference hardware, exposing them through a REST API that's a near drop-in replacement for the OpenAI chat completions endpoint. Pricing is pay-per-token and varies by model (roughly $0.05-$3 per million tokens), with a free API key available to get started and no separate infrastructure fee. The console also offers batch processing at a 50% discount and prompt-caching discounts for repeated context.

Not Human Search indexes websites and MCP servers by "agentic readiness" — how easily an AI agent can interact with a site via APIs, machine-readable docs, or MCP — exposing search, per-domain reports, and a live /check crawl that returns a 7-signal readiness score. It ships a public OpenAPI spec and an llms.txt file at its API root, with anonymous access to basic queries capped at a free 10-checks-per-hour rate limit, and Stripe-billed starter/pro/scale plans for higher-volume access or commerce endpoints like score monitoring and GEO score-fix services.