Across 489 live text-generation models, the major providers sit startlingly far apart on price. We compared the 20 providers with a real model lineup.
On a typical 3:1 output-to-input blend, Meta averages about $0.26 per 1M tokens across its 12 models, while OpenAI averages $27.33 across 80 — roughly 107× more. Same task, very different bill, depending only on who you call.
Premium providers price for frontier capability, large context and ecosystem; volume-and-efficiency providers price to win developers and high-throughput workloads. Neither is wrong — they target different buyers. The mistake is assuming the expensive option is automatically the better fit.
Start from the job, not the brand. For routing, extraction, drafting and most high-volume work, a low-cost provider often matches the result at a fraction of the cost; reserve premium models for the hardest reasoning. The map below plots every model by price and efficiency — up-and-to-the-left is better value.
Every tracked model plotted by input price (log scale) and composite efficiency. Toward the top-left means better value per dollar.
Each dot is one model · color = region · click a dot to open it.
Pricing is real (via OpenRouter, updated daily). This is market analysis, not investment or procurement advice.