The price floor for usable text generation has collapsed. Across 489 live generation models, we mapped just how low — and how crowded — the bottom of the market has become.
The cheapest generation model we track, Llama-3.2-3B-Instruct, runs about $0.020 per 1M tokens on a 3:1 blend. That is not a typo — small, efficient models have pushed the floor close to zero, and quality at the low end keeps rising.
234 of 489 generation models now cost under $1 per 1M tokens, and 161 come in under $0.50. Meanwhile the most expensive tenth of the catalogue averages $49.94. For high-volume, price-sensitive work, the cheap tier is no longer a compromise — it is the default.
Rock-bottom prices usually mean smaller models, shorter context or fewer modalities — fine for routing, classification, extraction and bulk drafting, less so for the hardest reasoning. Match the model to the job: cheap where you can, premium only where it measurably pays. The map below plots price against efficiency.
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.