Multimodal · Benchmark profile
A referring-expression grounding benchmark averaged across RefCOCO variants to test whether a model can localize described objects correctly.
Data verified 21 Jul 2026 · Methodology 1.6.0
Benchmark score on RefCOCO average
Qwen3.6 27B leads at 92.5%, followed by Qwen3.6-35B-A3B (92%) and Nemotron 3 Nano Omni 30B A3B (90.5%).
Visual analysis
Switch between model placement, score distribution and descriptive provider averages. Every view uses the same sourced leaderboard.
4 ranked models · higher is better
| Model | Rank | Provider | Score |
|---|---|---|---|
| Qwen3.6 27B | #1 | Alibaba Cloud | 92.5% |
| Qwen3.6-35B-A3B | #2 | Alibaba Cloud | 92% |
| Nemotron 3 Nano Omni 30B A3B | #3 | NVIDIA | 90.5% |
| Interfaze Beta | #4 | Interfaze | 82.1% |
One best score per model · higher is better
| Rank | Model | Provider | License | Evidence use | Score |
|---|---|---|---|---|---|
| #1 | Qwen3.6 27B qwen3.6-27b | Alibaba Cloud | open | Estimated reference | 92.5% |
| #2 | Qwen3.6-35B-A3B qwen3-6-35b-a3b | Alibaba Cloud | open | Estimated reference | 92% |
| #3 | Nemotron 3 Nano Omni 30B A3B nemotron-3-nano-omni-30b-a3b | NVIDIA | open | Estimated reference | 90.5% |
| #4 | Interfaze Beta interfaze-beta | Interfaze | closed | Estimated reference | 82.1% |
The top of this snapshot is led by Qwen3.6 27B at 92.5%; third place is 2.0 points behind. The top-4 spread is 10.4 points.
About RefCOCO average
A referring-expression grounding benchmark averaged across RefCOCO variants to test whether a model can localize described objects correctly. Results stay tied to the exact model variant and evaluation system. Multiple systems for the same model use the best published score on this page; overall Lumina scoring uses the median of ranking-eligible rows.
Open benchmark source ↗FAQ
A referring-expression grounding benchmark averaged across RefCOCO variants to test whether a model can localize described objects correctly.
Qwen3.6 27B by Alibaba Cloud currently leads with 92.5%.
4 models in the LuminaBench cohort have a qualifying score on this benchmark.
No. This benchmark is display-only and does not enter the overall Lumina composite.
Related
