Google launches Nano Banana 2.1 image model at about half the price

The headline number is price, but the more consequential detail is provenance: the performance claims rest entirely on Google's own testing, with no independent benchmarks cited. That matters in a market where image-model buyers increasingly weigh cost against verifiable capability, and where self-reported gains have a poor track record of surviving outside scrutiny. Halving cost lowers the barrier for high-volume generation, but the open question is whether third-party evaluations emerge to test the performance claims — and whether rivals respond on price or on measured quality.