PromptBattleLab

Image model library

Every text-to-image model, one honest index.

Still-image models we track and battle-test with identical prompts — separate from the video model library. Specs from official docs, prices from real fal.ai API bills. Updated July 2026.

Image model battles

Battle-tested — july 2026

ModelVendorModeMax resAPI accessPrice / run
Nano Banana 2Googlet2i1K–2Kfal.ai, Gemini API~$0.08/image @fal
Nano Banana ProGooglet2i1K–2Kfal.ai, Gemini API~$0.15/image @fal
FLUX.2 ProBlack Forest Labst2iup to 2Kfal.ai, BFL API~$0.05/image @fal
Seedream 5.0 LiteByteDancet2iup to 2Kfal.ai~$0.035/image @fal
Seedream 5.0 ProByteDancet2iup to 2Kfal.ai~$0.0675/image @fal

Imagen 4

active

Google's production image family — on the list for a future battle once we lock the same fal vs direct API comparison workflow.

queued for battle

GPT Image

active

Strong prompt adherence and text-in-image — not yet run through our identical-prompt battle pipeline.

queued for battle

Ideogram 3

active

Typography and layout leader — queued for a poster-style image battle, not this macro product test.

queued for battle

Video models live on the video model library. Image battles publish rankings, per-model notes, and outputs at the bottom of each comparison page — see how we test.