PerceptionBench with tools (PerceptionBench (tools))
We show this table for reference; we do not rank on it.
PerceptionBench fine-grained visual perception tasks run with tool use, as reported by Meta.
Benchmark score on PerceptionBench (tools) — November 11, 2026
We compile the PerceptionBench (tools) rows from provider self-reports. Muse Ember leads the table at 80.6%, followed by Claude Haiku 5.5 (78.6%) and GPT-6 Luna (62.8%). We do not use these results to rank models overall.
Muse Ember
Meta
Claude Haiku 5.5
Anthropic
GPT-6 Luna
OpenAI
3 modelsMultimodal & GroundedCurrentDisplay onlyUpdated November 11, 2026
Benchmark score table (3 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | Muse EmberMeta | 80.6% | Not reported | Closed |
| 2 | Claude Haiku 5.5Anthropic | 78.6% | Not reported | Closed |
| 3 | GPT-6 LunaOpenAI | 62.8% | Not reported | Closed |
Among the reported PerceptionBench (tools) rows, Muse Ember is first at 80.6%. The third row is 17.8 points behind. The broader top-10 range is 17.8 points, so the table still separates the published systems.
3 models have been evaluated on PerceptionBench (tools). The benchmark falls in the Multimodal & Grounded category. PerceptionBench (tools) is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About PerceptionBench (tools)
Year
2026
Tasks
Fine-grained visual-perception tasks with tools
Format
Provider-reported score
Difficulty
Fine-grained visual perception
Meta reports Muse Ember and its comparison models on PerceptionBench with tools in its internal evaluation suite, at max reasoning effort. BenchLM keeps tool-assisted runs apart from the tool-free PerceptionBench values that Moonshot and others report, and stores them display-only.
Freshness and provenance
Version
PerceptionBench (tools) 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does PerceptionBench (tools) measure?
PerceptionBench fine-grained visual perception tasks run with tool use, as reported by Meta.
Which model scores highest on PerceptionBench (tools)?
Muse Ember by Meta currently leads with a score of 80.6% on PerceptionBench (tools).
How many models are evaluated on PerceptionBench (tools)?
3 AI models have published results on PerceptionBench (tools) in the BenchLM catalog.
Compare top models on PerceptionBench (tools)
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 7,000+ subscribers.
One email each week. Unsubscribe anytime.