ODINW13
We show this table for reference; we do not rank on it.
A visual detection and grounding benchmark slice used to compare zero-shot object understanding across diverse domains.
Benchmark score on ODINW13 — October 17, 2026
We compile the ODINW13 rows from provider self-reports and secondary reports. Qwen3.7 Plus leads the table at 51.1%, followed by Qwen3.6-35B-A3B (50.8%) and Qwen3.5 397B (47.0%). We do not use these results to rank models overall.
Qwen3.7 Plus
Alibaba
Qwen3.6-35B-A3B
Alibaba
Qwen3.5 397B
Alibaba
4 modelsMultimodal & GroundedCurrentDisplay onlyUpdated October 17, 2026
Benchmark score table (4 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | Qwen3.7 PlusAlibaba | 51.1% | Not reported | Closed |
| 2 | Qwen3.6-35B-A3BAlibaba | 50.8% | Not reported | Open |
| 3 | Qwen3.5 397BAlibaba | 47.0% | Not reported | Open |
| 4 | Gemini 3 ProGoogle | 46.3% | Not reported | Closed |
Among the reported ODINW13 rows, Qwen3.7 Plus is first at 51.1%. The third row is 4.1 points behind. The broader top-10 range is 4.8 points, so many of the published results sit in a relatively narrow band.
4 models have been evaluated on ODINW13. The benchmark falls in the Multimodal & Grounded category. ODINW13 is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About ODINW13
Year
2026
Tasks
Out-of-distribution object understanding
Format
Detection and grounding
Difficulty
Robust visual grounding
ODINW-style scores matter because real deployment images often differ sharply from canonical internet photos. This benchmark checks whether object understanding transfers out of distribution.
Freshness and provenance
Version
ODINW13 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 ODINW13 measure?
A visual detection and grounding benchmark slice used to compare zero-shot object understanding across diverse domains.
Which model scores highest on ODINW13?
Qwen3.7 Plus by Alibaba currently leads with a score of 51.1% on ODINW13.
How many models are evaluated on ODINW13?
4 AI models have published results on ODINW13 in the BenchLM catalog.
Compare top models on ODINW13
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 5,600+ readers.
One email each week. Unsubscribe anytime.