# PerceptionBench leaderboard

Multimodal & Grounded benchmark

> Every figure below is reproduced as its upstream source published it: nothing is modelled, estimated, interpolated or converted. `Unavailable` means no source published the value — it is never a zero. Each value carries its evidence state and the date it was observed.

Page: https://modelscale.dev/leaderboards/benchmarks/multimodal-grounded-perception-bench

PerceptionBench (Internal). Every model the catalog carries a published PerceptionBench value for, ranked by that value.

## Benchmark record

| Field | Value |
| --- | --- |
| Category | Multimodal & Grounded |
| Measure | Internal evaluation score |
| Tasks | Internal atomic visual-perception tasks |
| Difficulty | Fine-grained visual perception |
| Published by | [Kimi K3: Open Frontier Intelligence](https://www.kimi.com/blog/kimi-k3) |

Moonshot AI's internal benchmark for atomic visual perception capabilities.

## PerceptionBench ranking

6 models with a published PerceptionBench value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.

| Rank | Model | Provider | Internal evaluation score | Evidence |
| --- | --- | --- | --- | --- |
| 1 | [Kimi K3.1 Pro](https://modelscale.dev/models/kimi-3-1-pro) | Moonshot AI | 68.3 | Observed 2026-11-21 · source benchlm:models |
| 2 | [Inkling 2 Small](https://modelscale.dev/models/inkling-2-small) | Thinking Machines Lab | 64.7 | Observed 2026-11-21 · source benchlm:models |
| 3 | [Qwen3.8 Max](https://modelscale.dev/models/qwen3-8-max) | Alibaba | 63.5 | Observed 2026-11-21 · source benchlm:models |
| 4 | [Kimi K3](https://modelscale.dev/models/kimi-3) | Moonshot AI | 58.5 | Observed 2026-11-21 · source benchlm:models |
| 5 | [Inkling 2 Mini](https://modelscale.dev/models/inkling-2-mini) | Thinking Machines Lab | 57.5 | Observed 2026-11-21 · source benchlm:models |
| 6 | [dots3-note Preview](https://modelscale.dev/models/dots3-note-preview) | Dots Studio | 53.4 | Observed 2026-11-21 · source benchlm:models |

Rows are ordered by the value the source published, highest first. The source does not state whether a higher value is the better result, so this document does not either: for a benchmark that measures a rate of failure, read the table from the bottom.
