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.

Spatial reasoning

Objects are positioned, ordered, and oriented exactly as specified (left/right, on-top-of, behind, counts).

  1. 1

    GPT Image 1.5

    8.5

  2. 2

    Qwen Image Generator with LoRA

    8.3

  3. 3

    FLUX.2 Max Image Generator

    8.0

  4. 4

    FLUX.2 Pro Image Generator

    8.0

  5. 5

    Nano Banana Pro Image Generator

    8.0

  6. 6

    Seedream V4.5

    8.0

  7. 7

    FLUX.2 Flex Image Generator

    7.0

  8. 8

    FLUX.2 Turbo Image Generator

    7.0

  9. 9

    FLUX.1 Schnell Image Generator

    6.3

  10. 10

    FLUX.1 Dev Image Generator

    6.0

  11. 11

    FLUX.2 Klein 9B Image Generator

    4.0