LLM API pricing in 2026, compared across Claude, GPT, Gemini, and Kimi

tldr: Output tokens cost five to eight times input tokens at every vendor, so your output-to-input ratio decides your bill more than any headline rate does. Current verified rates are below. Two rules move real bills more than the rate card: Anthropic’s newer models produce about 30% more tokens for the same text, and both OpenAI and Google surcharge long-context requests while Anthropic does not.