Trending AI image research daily
Every major text-to-image model, ranked by Elo from blind head-to-head human votes — no cherry-picked grids, no lab-supplied samples. Filter by prompt category, flip on the open-weights cut, and see where the checkpoints you can download stand against the closed APIs.
45
Models ranked
8
Open weights
313,593
Arena appearances
5
Prompt categories
Arena Elo
Higher is betterOverall leaderboard · best model per lab
- 1GPT Image 2 (high)1337
- 2Reve 2.01273
- 3MAI-Image-2.51266
- 4HiDream-O1-Image-1.51263
- 5Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)1257
- 6Riverflow 2.01249
- 7Cosmos3-Super-Text2Image (agentic)1219
- 8Recraft V4.1 Utility Pro1206
- 9grok-imagine-image-quality1201
Open Weights
Higher is betterDownloadable checkpoints · best per lab
- 1Cosmos3-Super-Text2Image (agentic)1219
- 2HiDream-O1-Image-Dev-26041183
- 3Ideogram 4.0 Quality1166
- 4ERNIE Image1163
- 5Qwen Image Max 25121154
- 6FLUX.2 [dev]1153
- 7FLUX.2 [dev] Turbo1151
Price per Image
Lower is betterUSD per generation, default quality · cheapest per lab
- 1FLUX.2 [dev] Turbo$0.01
- 2ERNIE Image Turbo$0.01
- 3FLUX.2 [dev]$0.01
- 4MAI-Image-2.5-Flash$0.02
- 5grok-imagine-image$0.02
- 6Qwen Image Max 2512$0.02
- 7Reve 2.0$0.02
- 8Krea 2 Medium$0.03
- 9Seedream 4.0$0.03
Global leaderboard
Elo from blind head-to-head votes · All categories
| Rank | Model | Creator | Elo | 95% CI | Appearances | $/Image | Released |
|---|---|---|---|---|---|---|---|
| 1 | GPT Image 2 (high) | OpenAI | 1337 | ±8 | 13,921 | $0.21 | Apr 2026 |
| 2 | Reve 2.0 | Reve | 1273 | ±11 | 3,082 | $0.02 | Jun 2026 |
| 3 | MAI-Image-2.5 | Microsoft AI | 1266 | ±10 | 6,079 | $0.05 | Jun 2026 |
| 4 | HiDream-O1-Image-1.5 | HiDream | 1263 | ±9 | 6,601 | $0.08 | Jun 2026 |
| 5 | GPT Image 1.5 (high) | OpenAI | 1257 | ±8 | 9,158 | $0.13 | Dec 2025 |
| 6 | Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image) | 1257 | ±11 | 6,125 | — | Jun 2026 | |
| 7 | Nano Banana 2 (Gemini 3.1 Flash Image Preview) | 1252 | ±8 | 11,454 | $0.07 | Feb 2026 | |
| 8 | Riverflow 2.0 | Sourceful | 1249 | ±9 | 7,092 | $0.15 | Feb 2026 |
| 9 | Cosmos3-Super-Text2Image (agentic)Open | NVIDIA | 1219 | ±9 | 6,710 | — | May 2026 |
| 10 | Nano Banana Pro (Gemini 3 Pro Image) | 1217 | ±8 | 8,551 | $0.13 | Nov 2025 | |
| 11 | Recraft V4.1 Utility Pro | Recraft | 1206 | ±9 | 6,841 | $0.21 | May 2026 |
| 12 | MAI-Image-2.5-Flash | Microsoft AI | 1205 | ±10 | 3,999 | $0.02 | Jun 2026 |
| 13 | grok-imagine-image-quality | xAI | 1201 | ±8 | 7,597 | $0.05 | Apr 2026 |
| 14 | Recraft V4.1 Utility | Recraft | 1196 | ±9 | 4,567 | $0.04 | May 2026 |
| 15 | Krea 2 Medium | Krea | 1195 | ±9 | 6,793 | $0.03 | May 2026 |
| 16 | FLUX.2 [max] | Black Forest Labs | 1192 | ±8 | 8,389 | $0.07 | Dec 2025 |
| 17 | MAI-Image-2 | Microsoft AI | 1187 | ±8 | 6,826 | $0.04 | Mar 2026 |
| 18 | Krea 2 Large | Krea | 1186 | ±9 | 7,132 | $0.06 | May 2026 |
| 19 | Seedream 4.0 | ByteDance Seed | 1186 | ±7 | 14,140 | $0.03 | Sep 2025 |
| 20 | FLUX.2 [pro] | Black Forest Labs | 1185 | ±8 | 7,948 | $0.03 | Nov 2025 |
| 21 | HiDream-O1-Image-Dev-2604Open | HiDream | 1183 | ±8 | 6,260 | — | May 2026 |
| 22 | Imagen 4 Ultra Preview 0606 | 1178 | ±8 | 5,413 | $0.06 | Jun 2025 | |
| 23 | FLUX.2 [flex] | Black Forest Labs | 1177 | ±8 | 5,712 | $0.06 | Nov 2025 |
| 24 | grok-imagine-image | xAI | 1174 | ±8 | 9,022 | $0.02 | Jan 2026 |
| 25 | Luma UNI 1 Max | Luma Labs | 1172 | ±9 | 4,894 | $0.10 | May 2026 |
| 26 | Imagen 4 Ultra | 1170 | ±8 | 6,105 | $0.06 | Jun 2025 | |
| 27 | Qwen Image 2.0 Pro (2026-04-22) | Alibaba | 1170 | ±9 | 4,970 | $0.07 | Apr 2026 |
| 28 | ImagineArt 2.0 | ImagineArt | 1168 | ±8 | 10,045 | $0.03 | Apr 2026 |
| 29 | Ideogram 4.0 QualityOpen | Ideogram | 1166 | ±9 | 4,015 | $0.06 | Jun 2026 |
| 30 | ERNIE ImageOpen | Baidu | 1163 | ±9 | 3,971 | $0.03 | Apr 2026 |
| 31 | Seedream 4.5 | ByteDance Seed | 1163 | ±7 | 9,993 | $0.04 | Dec 2025 |
| 32 | ERNIE Image TurboOpen | Baidu | 1160 | ±9 | 4,933 | $0.01 | Apr 2026 |
| 33 | Imagen 4 Preview 0606 | 1160 | ±8 | 5,388 | $0.04 | Jun 2025 | |
| 34 | MAI-Image-2-Efficient | Microsoft AI | 1156 | ±8 | 7,541 | $0.02 | Apr 2026 |
| 35 | Qwen Image Max 2512Open | Alibaba | 1154 | ±8 | 5,192 | $0.02 | Dec 2025 |
| 36 | Recraft V4.1 Pro | Recraft | 1154 | ±9 | 4,295 | $0.21 | May 2026 |
| 37 | Nano Banana (Gemini 2.5 Flash Image) | 1153 | ±8 | 10,728 | $0.04 | Aug 2025 | |
| 38 | FLUX.2 [dev]Open | Black Forest Labs | 1153 | ±8 | 7,096 | $0.01 | Nov 2025 |
| 39 | FLUX.2 [dev] TurboOpen | Fal | 1151 | ±8 | 5,791 | $0.01 | Dec 2025 |
| 40 | Recraft V4.1 | Recraft | 1148 | ±9 | 4,625 | $0.04 | May 2026 |
| 41 | ImagineArt 1.5 Preview | ImagineArt | 1147 | ±8 | 9,576 | $0.03 | Nov 2025 |
| 42 | Luma UNI 1 | Luma Labs | 1145 | ±9 | 5,441 | $0.04 | May 2026 |
| 43 | grok-imagine-image-pro | xAI | 1144 | ±8 | 8,152 | $0.07 | Feb 2026 |
| 44 | Seedream 3.0 | ByteDance Seed | 1143 | ±8 | 6,046 | $0.03 | Apr 2025 |
| 45 | Wan 2.5 Preview | Alibaba | 1142 | ±8 | 5,384 | $0.02 | Sep 2025 |
Showing 45 of 45 models. Category Elo is computed from votes on that prompt type alone.
Arena Elo — Open Weights
Every downloadable checkpoint on the board, ranked by overall Elo. Higher is better.
Cosmos3 sits fifth overall — ahead of most closed APIs — and the FLUX.2 [dev] line keeps Black Forest Labs in four of the fourteen open slots.
Arena Elo vs. Price per Image
Crowd-judged quality against USD per generation (log scale).
Up and to the left wins. The open-weights mid-table — FLUX.2 [dev], Qwen-Image-2512, Hunyuan — is where the value lives; models without public per-image pricing are excluded.
Frontier Elo Over Time
Arena Elo by release date. The dashed line tracks the running frontier.
The frontier has changed hands five times since Seedream 3.0 — ByteDance Seed, Google, and OpenAI have all held it. GPT Image 2 set the current mark in April 2026.
How the arena works
1 · Prompt
Prompts are sampled across five categories — people, photorealism, art, product, and typography — so no model can farm a single style.
2 · Blind matchup
Two models generate from the same prompt. Voters see the images side by side with no names, no logos, and no way to tell who made what.
3 · Elo
Each vote updates both models' Elo ratings. Confidence intervals come from resampling the vote history; new models stay provisional until they clear roughly 3,000 appearances.
FAQ
Which model leads the arena right now?
GPT Image 2 (high) holds the top spot with an Elo of 1337 across 13,921 appearances — a 64-point gap over the chasing pack.
And among open weights?
NVIDIA's Cosmos3-Super-Text2Image (agentic) leads the open-weights cut at 1219 — ninth overall, ahead of plenty of closed APIs — followed by HiDream-O1-Image-Dev-2604 (1183) and Ideogram's Ideogram 4.0 Quality (1166).
How are models ranked?
Pure crowd judgment. Voters get a prompt and two unlabeled images from different models and pick the better one; thousands of those blind matchups feed an Elo rating, with a 95% confidence interval from resampling the vote history.
What counts as open weights?
The checkpoint has to be downloadable and runnable on your own hardware under a published license. API-only siblings of open families — FLUX.2 [pro], Qwen Image 2.0 Pro — stay in the proprietary column.
Latest Research
Reporting from the paper mines

Study Shows Private Label Products Win Big in AI Search Visibility
By Store Brands · 2 hrs ago

AI Companions Pose Hidden Risks To Child Development, Expert Warns In New Book
By Foreign Policy Journal · 1 hr ago

Government must stand up to global AI giants - AdNews
By adnews.com.au · 55 mins ago

Headcount 'the laziest way' to judge agency capability - AdNews
By adnews.com.au · 55 mins ago
Methodology: rankings come from blind pairwise human votes; Elo, confidence intervals, and appearance counts reflect the full vote history at snapshot time. Arena data: public image-arena snapshot, July 15, 2026. Prices are list API prices per image at default quality and change frequently.


