AICost is an independent price desk for AI models that answers one practical question: what will a given model actually cost to run. It tracks input, output and cache rates per million tokens for more than 340 models from 35 providers, including OpenAI, Anthropic, Google, Mistral, DeepSeek, Qwen and Kimi, with every rate read from the vendor’s own price page and stamped with the date it was read. Around that pricing hub sit head-to-head comparisons with live pricing, benchmarks and verdicts, capability questions answered with verified test outputs and screenshots, and nine live calculators covering LLM API cost, token counting, subscription versus API, video generation, text-to-speech, embeddings, fine-tuning, agent cost and image generation. A glossary ties AI terms to what they cost, and a journal logs price moves, model releases and policy changes within hours. The operator verifies each claim against the vendor’s canonical page and keeps a public correction log when a number is fixed. It is a free reference for developers, founders and finance-minded builders who choose models on cost as well as capability and want the arithmetic checked against the source, not a press release.