ABACUS
Models add, remove, or change a specified number of objects in an image. The evaluator checks the count and preservation of the surrounding scene.
- Made by
- Aymen Lassoued and colleagues
- Released
- October 2026
- Kind
- Image generation and editing
- Level
- School
- Scored
- Checked automatically
- Answers
- Open-ended
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Single turn
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Vision, Image, video and 3D generation
- Scores refreshed
- November 20th, 2026
Seedream 4.0 leads with 0.31
Best score per model from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best.
- 1
Seedream 4.0 0.31
- 2
Gemini 2.5 Flash Image Preview (nano-banana) 0.29
- 3 GPT-Image-1 0.26
- 4
Qwen-Image-Edit 0.16
- 5 FLUX.1 Kontext [dev] 0.14
- 6
Step1X-Edit 0.11
- 7
BAGEL 0.09
Source: Official ABACUS leaderboard
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 0.31 | Official leaderboard | ||
| 2 | | 0.29 | Official leaderboard | ||
| 3 | GPT-Image-1 | 0.26 | Official leaderboard | ||
| 4 | | 0.16 | Official leaderboard | ||
| 5 | FLUX.1 Kontext [dev] | 0.14 | Official leaderboard | ||
| 6 | | 0.11 | Official leaderboard | ||
| 7 | | 0.09 | Official leaderboard |
Rankings by part
We read 9 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Add: Easy, Add: Medium and Add: Hard. Seedream 4.0 leads every one of them. Arrows show how far a model moved from its overall rank among the models we track.
Add: Easy
7 models we track have a score here.
- 1
Seedream 4.0 0.55
- 2
Gemini 2.5 Flash Image Preview (nano-banana) 0.52
- 3 GPT-Image-1 0.51
- 4
Qwen-Image-Edit 0.35
- 5 FLUX.1 Kontext [dev] 0.30
- 6
Step1X-Edit 0.24
- 7
BAGEL 0.20
Add: Medium
7 models we track have a score here.
- 1
Seedream 4.0 0.42
- 2
Gemini 2.5 Flash Image Preview (nano-banana) 0.39
- 3 GPT-Image-1 0.36
- 4
Qwen-Image-Edit 0.23
- 5 FLUX.1 Kontext [dev] 0.20
- 6
Step1X-Edit 0.16
- 7
BAGEL 0.13
Add: Hard
7 models we track have a score here.
- 1
Seedream 4.0 0.28
- 2
Gemini 2.5 Flash Image Preview (nano-banana) 0.26
- 3 GPT-Image-1 0.25
- 4
Qwen-Image-Edit 0.13
- 5 FLUX.1 Kontext [dev] 0.12
- 6
Step1X-Edit 0.10
- 7
BAGEL 0.08
Remove: Easy
7 models we track have a score here.
- 1
Seedream 4.0 0.46
- 2
Gemini 2.5 Flash Image Preview (nano-banana) 0.45
- 3 GPT-Image-1 0.38
- 4
Qwen-Image-Edit 0.26
- 5 FLUX.1 Kontext [dev] 0.24
- 6
Step1X-Edit 0.18
- 7
BAGEL 0.14
Remove: Medium
7 models we track have a score here.
- 1
Seedream 4.0 0.31
- 2
Gemini 2.5 Flash Image Preview (nano-banana) 0.30
- 3 GPT-Image-1 0.24
- 4
Qwen-Image-Edit 0.18
- 5 FLUX.1 Kontext [dev] 0.14
- 6
Step1X-Edit 0.12
- 7
BAGEL 0.08
Remove: Hard
7 models we track have a score here.
- 1
Seedream 4.0 0.20
- 2
Gemini 2.5 Flash Image Preview (nano-banana) 0.18
- 3 GPT-Image-1 0.17
- 4 FLUX.1 Kontext [dev] ↑1 0.08
- 5
Qwen-Image-Edit ↓1 0.08
- 6
Step1X-Edit 0.07
- 7
BAGEL 0.05
Change: Easy
7 models we track have a score here.
- 1
Seedream 4.0 0.38
- 2
Gemini 2.5 Flash Image Preview (nano-banana) 0.31
- 3 GPT-Image-1 0.31
- 4
Qwen-Image-Edit 0.18
- 5 FLUX.1 Kontext [dev] 0.14
- 6
Step1X-Edit 0.11
- 7
BAGEL 0.10
Change: Medium
7 models we track have a score here.
- 1
Seedream 4.0 0.25
- 2
Gemini 2.5 Flash Image Preview (nano-banana) 0.22
- 3 GPT-Image-1 0.18
- 4
Qwen-Image-Edit 0.11
- 5 FLUX.1 Kontext [dev] 0.09
- 6
BAGEL ↑1 0.06
- 7
Step1X-Edit ↓1 0.06
Change: Hard
7 models we track have a score here.
- 1
Seedream 4.0 0.14
- 2
Gemini 2.5 Flash Image Preview (nano-banana) 0.12
- 3 GPT-Image-1 0.11
- 4 FLUX.1 Kontext [dev] ↑1 0.05
- 5
Qwen-Image-Edit ↓1 0.05
- 6
BAGEL ↑1 0.03
- 7
Step1X-Edit ↓1 0.03
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
No dated values yet.
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Vision
- MMMU In use 11.5K college-level multimodal questions across 30 subjects. Multiple-choice and open accuracy.
Qwen 3.6 Plus 86%
- MMMU-Pro In use Harder MMMU: college-level multimodal questions with more options and vision-only inputs to remove text shortcuts. GPT-5.6 Sol 84.6%
- CharXiv In use Chart understanding on real arXiv figures. Reasoning questions require synthesizing information from complex charts. Claude Fable 5 93.5%
- MathVista In use Math reasoning in visual contexts (charts, geometry, puzzles) from 31 datasets. Answer accuracy.
Seed2.1 Pro 90.7%
- MATH-Vision (MathVision) In use 3,040 competition math problems with visual context. Answer accuracy.
Qwen 3.8 Max 97.7%
- MMStar In use Models answer 1,500 human-curated image questions that cannot be solved without the image, across 6 core capabilities. Scored by accuracy.
Qwen 3.8 Max 85.9%
Data audit
All 8 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
- Pass Independent runs Scores from the official leaderboard (7 entries).
- Pass Models scored 7 models we track.
- Note Newest score No scores yet.
- Note Frontier models None of its models are on the capabilities index.
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on cea-list.github.io: 7 rows for models we track at the last read.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.