Artem Voloskovets
Google just pulled ahead in the text-to-image race. Alphabet's (GOOGL) Nano Banana 2, publicly known as the Gemini 3.1 Flash Image Preview, has taken the #1 spot on the Artificial Analysis Text-to-Image leaderboard — edging out OpenAI and Google's own higher-end model in the process. For developers and enterprises building image-heavy products, this matters for one simple reason: you're getting top-tier output without paying top-tier prices.
Nano Banana 2 Scores ELO 1,272 Across 2,721 Samples
Nano Banana 2 achieved an ELO score of 1,272, evaluated across 2,721 samples — enough to land it firmly in first place. OpenAI's GPT Image 1.5 (high) follows close behind at 1,268, while Google's own Nano Banana Pro (Gemini 3 Pro Image) ranks third at 1,220. Other notable entries include Black Forest Labs' FLUX.2 variants and ByteDance Seed's Seedream series, but none managed to close the gap at the top.
In Image Editing, Nano Banana 2 placed #3, trailing only GPT Image 1.5 and Nano Banana Pro — a strong showing across both categories that confirms this isn't a one-dimensional result.
$67 Per 1,000 Images vs. $133-$140 for Competitors
The pricing gap is where things get interesting. Nano Banana 2 runs at approximately $67 per 1,000 images — about half the cost of Nano Banana Pro at $134 and OpenAI's GPT Image 1.5 at $133. FLUX.2 [max] for editing comes in even higher at $140 per 1,000. That cost differential gives Nano Banana 2 a compelling edge for any team running high-volume image generation. You can learn more about Google's broader model strategy in coverage of how OpenAI faces new competition as Nano Banana 2 and Gemini 3.0 launched.
The model is available through both the Gemini App and the Gemini API, making it accessible to a wide range of users from solo developers to enterprise teams. Notably, while Nano Banana 2 leads on image generation, Google's larger models have faced separate performance scrutiny — Gemini 3.1 Pro throughput dropped 46% to 50 TPS on Google Vertex, a reminder that different models carry different trade-offs.
Still, for text-to-image use cases, the combination of benchmark leadership and aggressive pricing is hard to argue with. As explored in depth with Gemini 3.1 Pro powering 1M token workflows with NotebookLM, Google is clearly investing across its entire generative AI stack. In a market where cost efficiency increasingly drives platform choices, Nano Banana 2 looks set to shape adoption patterns well beyond this leaderboard cycle.
Artem Voloskovets
