o4-mini

By price, o4-mini ranks 41st cheapest of 68 on input and 42nd on output. Its quality score of 88 ranks 29th of 68, and on value (quality score ÷ output price) it ranks 39th of 68. Its blended rate sits 35% below the catalog average.

In real billing terms: a support chatbot handling 50,000 turns a month (2,000 input / 500 output tokens each) costs about $220 per month on o4-mini, while a RAG workload of 20,000 queries (8,000 input / 800 output tokens each) runs about $246. Because output is priced 4.0× input here, long-generation tasks scale cost fastest — cap output length before you optimise anything else.

A 200K context window holds roughly 267 pages of A4 text (at ~750 tokens per page), ranking 44th of 68 here. That covers typical RAG and multi-turn chat, though book-length inputs still need retrieval and chunking. Output runs at about 80 tok/s (47th fastest of 68), acceptable interactively, though long responses will feel slow.

We track 11 OpenAI models, and o4-mini's blended rate sits 50% below that provider's average. If your task tolerates a little less headroom, Hunyuan HY3 (Tencent) blends about 95% cheaper at a quality score of 87, and is worth benchmarking first. Going the other way, qwen3.5 scores 91 on quality for about 94% less.

The closest model in OpenAI's line-up is gpt-5.4-mini ($0.75 input / $4.50 output, quality 85). Against it, o4-mini scores 3 points higher on quality, runs 70 tok/s slower (about 47%), has a 849K smaller context window, costs 12% more blended. When two models sit this close on price, the deciding factor is usually the quality ceiling — run both against your real prompts before picking.