How much does GPT-4o cost per million tokens?
GPT-4o costs $2.50/million input tokens and $10 output; GPT-4o mini, $0.15/$0.60 (~17x cheaper). A request rarely costs more than $0.05. Available via the OpenAI API or a platform like Ask Aurel, no subscription needed.
GPT-4o (OpenAI) costs $2.50 per million input tokens and $10 per million output tokens in 2026; the GPT-4o mini variant is about 17 times cheaper, at $0.15 / $0.60.
(1) Volume benchmarks: 1 million tokens is roughly 750,000 words — about a 1,500-page book, a volume most individual users never reach.
(2) Cost per question: a typical conversation uses 1,000-5,000 tokens (prompt + reply), so a GPT-4o request rarely costs more than $0.05, and a GPT-4o mini request just a few thousandths of a dollar. For 10 questions a day on GPT-4o, expect $5-15 per month.
(3) Comparison with competitors: GPT-4o sits between Claude Sonnet 4.6 ($3/$15) and Gemini 2.5 Flash (roughly $0.30/$2.50), in the upper-mid range for a general-purpose multimodal model. GPT-4o mini, by contrast, is one of the most competitive rates on the market for simple, high-volume tasks.
(4) How to access it: directly via the OpenAI API (platform.openai.com) with an API key and prepaid credit system, or through aggregator platforms like Ask Aurel that combine GPT-4o with 5 other AIs (Claude, Gemini, Mistral, Grok, DeepSeek) on pay-as-you-go, no technical setup required. OpenAI pricing changes regularly — always check openai.com/api/pricing for the latest figures before any significant budget commitment.