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.
Product fidelity
Product (label, logo, shape, colors) is reproduced exactly; brand text stays correct.
1
Qwen Image Generator with LoRA
4 tests
6.3
2
FLUX.2 Klein 9B Image Generator
2 tests
5.5
3
FLUX.2 Max Image Generator
2 tests
5.3
4
FLUX Kontext Image Editor
4 tests
5.1
5
Nano Banana Pro Image Generator
1 test
5.0
6
FLUX.2 Pro Image Generator
2 tests
4.8
7
GPT Image 1.5
4 tests
4.5
8
FLUX.2 Flex Image Generator
2 tests
4.5
9
Seedream V4.5
1 test
4.0
10
FLUX.2 Turbo Image Generator
1 test
3.0