# CountBench 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://modelchoice.dev/leaderboards/benchmarks/multimodal-grounded-count-bench

Every model the catalog carries a published CountBench value for, ranked by that value.

## Benchmark record

| Field | Value |
| --- | --- |
| Category | Multimodal & Grounded |
| Measure | Image-grounded counting |
| Tasks | Visual counting tasks |
| Difficulty | Fine-grained visual perception |
| Published by | [Qwen3.6 launch benchmarks](https://qwen.ai/blog?id=qwen3.6) |

A visual counting benchmark that tests whether a model can count objects and entities reliably in complex scenes.

## CountBench ranking

10 models with a published CountBench 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 | Image-grounded counting | Evidence |
| --- | --- | --- | --- | --- |
| 1 | [Qwen3.6-27B](https://modelchoice.dev/models/qwen3-6-27b) | Alibaba | 97.8 | Observed 2026-11-21 · source benchlm:models |
| 2 | [Gemini 3 Pro](https://modelchoice.dev/models/gemini-3-pro) | Google | 97.3 | Observed 2026-11-21 · source benchlm:models |
| 3 | [Qwen3.5 397B](https://modelchoice.dev/models/qwen3-5-397b) | Alibaba | 97.2 | Observed 2026-11-21 · source benchlm:models |
| 4 | [Kimi K2.5](https://modelchoice.dev/models/kimi-k2-5) | Moonshot AI | 94.1 | Observed 2026-11-21 · source benchlm:models |
| 5 | [LFM3-VL-2B](https://modelchoice.dev/models/lfm3-vl-2b) | LiquidAI | 93.0 | Observed 2026-11-21 · source benchlm:models |
| 6 | [GPT-5.2](https://modelchoice.dev/models/gpt-5-2) | OpenAI | 91.9 | Observed 2026-11-21 · source benchlm:models |
| 7 | [Claude Opus 4.5](https://modelchoice.dev/models/claude-opus-4-5) | Anthropic | 90.6 | Observed 2026-11-21 · source benchlm:models |
| 8 | [ZAYA1-VL-8B](https://modelchoice.dev/models/zaya1-vl-8b) | Zyphra | 88.1 | Observed 2026-11-21 · source benchlm:models |
| 9 | [LFM2.5-VL-3B](https://modelchoice.dev/models/lfm2-5-vl-3b) | LiquidAI | 87.3 | Observed 2026-11-21 · source benchlm:models |
| 10 | [LFM2.5-VL-450M](https://modelchoice.dev/models/lfm2-5-vl-450m) | LiquidAI | 73.3 | 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.
