Model leaderboard

Every model we’ve run through the standard Glif tests, scored out of 10 by an LLM judge. Overall needs 3 judged tests to rank; pick a capability to rank on it alone.

Stylization

Faithfully hits a named style or era (e.g. 1920s rubber hose, tilt-shift miniature, psychedelic retro) without drifting to a generic look.

1
Grok Imagine Video
1 test
9.0
2
Image to Video (Kling O3 Pro)
1 test
9.0
3
Reference to Video (Kling O3 Pro)
1 test
9.0
4
Image to Video (Kling 2.5 Pro)
1 test
9.0
5
Image to Video (Veo 3.1)
1 test
9.0
6
FLUX.2 Flex Image Generator
2 tests
8.5
7
FLUX.2 Max Image Generator
2 tests
8.5
8
FLUX.2 Pro Image Generator
2 tests
8.5
9
FLUX.2 Turbo Image Generator
2 tests
8.5
10
Image to Video (Kling 2.6 Pro)
1 test
8.5
11
Nano Banana Pro Image Generator
2 tests
8.5
12
Sora 2
1 test
8.5
13
Seedance Pro 1.5
1 test
8.5
14
Seedream V4.5
2 tests
8.5
15
Image to Video (Vidu Q3)
1 test
8.5
16
GPT Image 1.5
6 tests
8.3
17
FLUX.1 Dev Image Generator
6 tests
8.2
18
FLUX.2 Klein 9B Image Generator
2 tests
8.0
19
Image to Video (Veo 3.1 Fast)
1 test
8.0
20
FLUX.1 Schnell Image Generator
6 tests
6.8
21
Qwen Image Generator with LoRA
6 tests
6.4