PerceptionBench B

AI benchmark · Multimodal

Built by
Moonshot AI
Tier
B why
Snapshot

Open benchmark ↗kimi.com/blog/perception-bench

What is PerceptionBench?

PerceptionBench measures atomic visual perception in multimodal LLMs across ten capabilities such as counting, OCR, localization and hallucination, isolated from reasoning and knowledge.

Who built PerceptionBench

Led by Moonshot AI (Kimi Team).

    Tier B

    B

    Established: 15 or more models scored, or curated RL directories rate it.

    How much stands behind a tier: usage, who backs it, a confirmed line from its own site or paper, and a clear read on X. High with three of the four, medium with two, low otherwise.

    • 21 models scored
    • in curated RL directories (2 pts)

    On X · unclear, low buzz, from 1 post about PerceptionBench in the last month

    Mentioned only once, as one of three evaluation benchmarks in a small account's paper on in-context example selection; there is no discussion of its quality.

    @karrsen0713 ↗

    How we got to tier B
    1. Adoption: 21 models scored. Tier B.
    2. On X: unclear, low buzz. “Mentioned only once, as one of three evaluation benchmarks in a small account's paper on in-context example selection; there is no discussion of its quality.” No move.
    3. In curated RL directories: 2 points, a floor of C.
    4. Confirmed: a line from its own site or paper places it.

    A clear view on X moves a tier one step up or down; most entries don't move. How tiers work.

    Where PerceptionBench sits

    Its domain on the map, live. Hover a logo for its name; click to open it. Open it on the full map

    The galaxy could not load. See every vendor on the board.

    Next: browse all benchmarks. Entry from the RL Research daily scrape of public sources, 2026-11-07 snapshot.