Ai Imagora

Nano Banana 2.1 Benchmark

How Ai Imagora evaluates Nano Banana 2.1 across common generation and editing stress tests — methodology first, no fabricated scores.

Ready to create? Generate with Nano Banana 2.1.

How we test

  • Same prompt text for a given category when comparing runs
  • Similar resolution settings where the demo allows
  • Manual review of artifacts (hands, text, edges, identity drift)
  • Separate passes for text-to-image vs upload-based editing
  • Notes dated when a suite is re-run after model or routing changes

Test categories

Each row is a checklist item. Publish Prompt → Image → Evaluation only after a real generation is saved with that prompt. Until then, use this page as the scoring rubric.

CategoryWhat we look forStatus
PortraitsSkin, lighting, identity stability across close framing.Rubric ready — run in demo
HandsFinger count and natural gesture failures under hard poses.Rubric ready — run in demo
Text renderingShort titles vs longer paragraphs on posters.Rubric ready — run in demo
Product photographyMaterial readability and catalog-style lighting.Rubric ready — run in demo
Character consistencySame outfit/hair cues across two prompts.Rubric ready — run in demo
Image editingBackground swap while preserving subject edges.Rubric ready — run in demo
Multi-image referencesUpload-guided edits staying close to the source.Rubric ready — run in demo
PhotorealismBelievable materials without plastic over-smoothing.Rubric ready — run in demo
Typography layoutsNegative space usable for marketing frames.Rubric ready — run in demo
Complex compositionsMultiple subjects and depth cues in one scene.Rubric ready — run in demo

Limitations

These tests are not a laboratory claim or a universal ranking. Results vary by seed, safety filters, prompt wording, and upload quality. We do not invent numeric leaderboards without published images.

Run your own suite

Use prompts from the prompt library, generate on the demo, and compare with Flux, Midjourney, or GPT Image.