Benchmark Almanac

ArtEdit-Bench-Lr

JarvisEvo 8.76 tops its leaderboard Gemini 2.5 Flash Image Preview (nano-banana) 8.62 leads the models we track
Not in model cards

Models retouch photos from English or Chinese instructions. GPT-4o scores instruction following and visual quality. Pixel errors measure fidelity to reference edits.

Adobe and KAIST independently ran the editing test. Both languages use the same 400 cases. At least 24 English cases reuse MMArt-Bench images and instructions.

Released
November 2025
Kind
Apps and MCP servers
Level
University
Scored
Graded by a model
Answers
Open-ended
Tasks
All public
Measures
Capability
Subject
Arts
Language
English, Chinese
In and out
Images, Text → Images
Interaction
Single turn
Items from
Another benchmark
Refresh
One fixed set
Run by
Anyone
Scores refreshed
November 20th, 2026
11
Entries in its paper, 5 models we track

JarvisEvo tops the leaderboard with 8.76 Gemini 2.5 Flash Image Preview (nano-banana) leads with 8.62

Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. The headline averages English and Chinese overall ratings, using the published bilingual rating when available. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is Gemini 2.5 Flash Image Preview (nano-banana), at 8.62. This view shows only models we track, each at its best.

  1. 1 JarvisEvo 8.76
  2. 2 Gemini 2.5 Flash Image Preview (nano-banana) 8.62
  3. 3 GPT-Image-1 8.27
  4. 4 UniWorld-v1 8.21
  5. 5 BAGEL 7.72
  6. 6 MonetGPT 7.71
  7. 7 FLUX.1 Kontext [dev] 7.30
  8. 8 Instruct-Pix2Pix 6.71
  9. 9 Step1X-Edit 4.84
  10. 10 MagicBrush 3.43
  11. 11 OmniGen 2.81

Source: The ArtEdit-Bench-Lr paper

#EntryScoreSettingDateSource
1 JarvisEvo open weights 8.76 Official leaderboard
2 Gemini 2.5 Flash Image Preview (nano-banana) 8.62 Official leaderboard
3 GPT-Image-1 8.27 Official leaderboard
4 UniWorld-v1 open weights 8.21 Official leaderboard
5 BAGEL 7.72 Official leaderboard
6 MonetGPT open weights 7.71 Official leaderboard
7 FLUX.1 Kontext [dev] 7.30 Official leaderboard
8 Instruct-Pix2Pix open weights 6.71 Official leaderboard
9 Step1X-Edit 4.84 Official leaderboard
10 MagicBrush open weights 3.43 Official leaderboard
11 OmniGen open weights 2.81 Official leaderboard
#ModelScoreSettingDateSource
1 Gemini 2.5 Flash Image Preview (nano-banana) 8.62 Official leaderboard
2 GPT-Image-1 8.27 Official leaderboard
3 BAGEL 7.72 Official leaderboard
4 FLUX.1 Kontext [dev] 7.30 Official leaderboard
5 Step1X-Edit 4.84 Official leaderboard

Rankings by part

We read 10 parts from the benchmark’s paper, each ranking models on one part of the test, such as English: L1, English: L2 and English: SC. Gemini 2.5 Flash Image Preview (nano-banana) leads 5 of them, and 2 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

English: L1

Reported L1 multiplied by 100. 5 models we track have a score here.

  1. 1 Gemini 2.5 Flash Image Preview (nano-banana) 11.5
  2. 2 Step1X-Edit ↑3 11.8
  3. 3 BAGEL 16.3
  4. 4 FLUX.1 Kontext [dev] 19.8
  5. 5 GPT-Image-1 ↓3 21.2

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 Tool use

All 265 in Tool use
  • AutomationBench In use Agent completes end-to-end business workflows inside a simulated company exposing dozens of REST APIs. Graded on resulting state. Gemini 4 Neon 56.3%
  • MCP Atlas In use Multi-step workflows over real MCP servers. Scores tool-use competency on realistic tasks. Gemini 4 Argon 90.7%
  • BFCL V4 In use Models call functions and tools, including multi-turn agentic tasks with web search and memory. Checked against expected calls and outcomes. Qwen3.7-Max 75%
  • τ³-bench Banking In use Agents act as banking customer-service reps over many turns with a simulated user, following policy. Pass^1 on the final database state. Qwen 3.8 Max 55.1%
  • Toolathlon-Verified In use 108 long-horizon tasks across 32 apps through about 600 MCP tools. Execution-based checkers inspect final environment state. GLM-5.3-Flash 78.4%
  • DeepSearchQA In use Questions that need multi-step search and synthesis. Final answers graded against references. Kimi K3 95%

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 benchmark’s paper (11 entries).
  • Pass Models scored 5 models we track, plus 6 other entries on its leaderboard.
  • 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 arxiv.org: 5 rows for models we track and 6 other entries 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.

How we check