Image Generation
Ranked by momentum — measured, not bought.
- 1Qwen Image↗ Rising
Alibaba's open-weight image model with strong multilingual text rendering and clean commercial licensing, runnable locally.
Score 100.00
Why it's risingQwen Image's engagement has increased 218% compared to approximately 7 days ago.
- 2GPT Image↗ Rising
OpenAI's image model and current quality leader, the first to win on both text accuracy and photorealism, used inside ChatGPT.
Score 43.08
Why it's risingGPT Image's engagement has increased 84% over the past week.
- 3Recraft↘ Cooling
Design-focused image tool that generates true editable vector (SVG) output, with brand-kit styles held consistent across assets.
Score 33.87
Why it's coolingRecraft's engagement has declined significantly, down 89% compared to the previous week.
- 4Nano BananaWatchlist — not enough signal yet
Google's image model, free in the Gemini app and the leader at photo editing, character consistency, and natural-language edits.
Score 30.72
What the data showsNano Banana has received 2 stories on Hacker News in the past 7 days with 24 total points and 7 comments.
- 5FLUXWatchlist — not enough signal yet
Black Forest Labs' photorealism leader, with camera-accurate detail, multi-reference consistency, and open weights on its Dev tier.
Score 27.00
What the data showsFLUX has accumulated 2557 stars and 179 forks on GitHub with its most recent update on March 12, 2026.
- 6MidjourneyWatchlist — not enough signal yet
Image generator prized for the most distinctive, art-directed aesthetic, favored for editorial and creative hero imagery.
Score 23.52
What the data showsMidjourney has generated 3 Hacker News stories with 10 total points and 2 comments over the past 7 days.
- 7IdeogramWatchlist — not enough signal yet
Image model specialized in typography, rendering in-image text at near-perfect accuracy for posters, logos, and marketing graphics.
Score 15.87
What the data showsIdeogram has limited recent visibility with minimal engagement on Hacker News and no public GitHub repository.
Not enough cross-source signal yet for a score or a trend. These are measured nightly — they move up the moment the data does.
- Seedream
ByteDance's image model with best-in-class text rendering, native 4K output, and strong multi-reference image consistency.
- Adobe Firefly
Adobe's commercially safe image model, trained on licensed content with enterprise IP indemnification, integrated across Creative Cloud.
- Reve
Approachable image generator with an element-by-element editing interface for refining specific parts of a result.