Categories › AI & Agents › Models, LLM APIs, routing, serving, training & GPUs
Daily prices, stock and measured LLM speeds for local AI computers and graphics cards.
Remote server: your requests and data go to localaiprices.com. Claude Code: claude mcp add --transport http local-ai-prices https://localaiprices.com/mcp · Claude app: Settings → Connectors → Add custom connector.
The tools this server listed when we connected to it, in its own words.
search_machines — Find local AI computers and graphics cards by budget, memory and type. Returns price, measured tok/s, value (tok/s per 1,000 of currency), wget_machine — One machine in full by id (from search_machines) or page URL: price per store with stock and check time, price history summary, every publisfind_for_model — Which tracked machines can run an open model (4-bit weights fit in memory), cheapest first, with measured tok/s where someone has published list_changes — Price drops, price rises, restocks and newly tracked machines, newest first. Set since to a timestamp and only later changes come back; reuslist_models — The open models we track: id, size, memory needed at 4-bit and 8-bit, how many machines fit each and how many have a measured speed, and theNot tested by KitScout yet. When our harness runs it, pros, cons and the tasks we tried appear here, with the date and evidence.