NOVA-63
A broad multilingual benchmark row from Qwen's launch comparisons intended to measure cross-lingual capability beyond a single language family.
Top models on NOVA-63 — November 21, 2026
As of November 21, 2026, Qwen3.5 397B leads the NOVA-63 leaderboard with 59.1% , followed by Qwen3.7 Max (59.0%) and Qwen3.7 Plus (58.8%).
Qwen3.5 397B
Alibaba
Reported by AlibabaQwen3.7 Max
Alibaba
Reported by AlibabaQwen3.7 Plus
Alibaba
Reported by Alibaba7 modelsMultilingual7% of Multilingual reference weightCurrentUpdated November 21, 2026
Leaderboard (7 models)
None of the 7 results on this page has a documented run by the benchmark or an independent evaluator: 4 reported by the labs that built the models and 3 reported in another publisher’s document.
| Rank | Model / configuration | Score | Who ran it | Parameters (B) | Open / closed |
|---|---|---|---|---|---|
| 1 | Qwen3.5 397BAlibabaReported by Alibaba | 59.1% | Reported by Alibaba | Not reported | Open |
| 2 | Qwen3.7 MaxAlibabaReported by Alibaba | 59.0% | Reported by Alibaba | Not reported | Closed |
| 3 | Qwen3.7 PlusAlibabaReported by Alibaba | 58.8% | Reported by Alibaba | Not reported | Closed |
| 4 | Qwen3.6 PlusAlibabaReported by Alibaba | 57.9% | Reported by Alibaba | Not reported | Closed |
| 5 | Claude Opus 4.5AnthropicAnother lab’s comparison table | 56.7% | Another lab’s comparison table | Not reported | Closed |
| 6 | Kimi K2.5Moonshot AIAnother lab’s comparison table | 56.0% | Another lab’s comparison table | Not reported | Open |
| 7 | GLM-5Z.AIAnother lab’s comparison table | 55.1% | Another lab’s comparison table | Not reported | Open |
- Lab-reported
- Benchmark-run
- Independent
According to BenchLM.ai, Qwen3.5 397B leads the NOVA-63 benchmark with a score of 59.1%, followed by Qwen3.7 Max (59.0%) and Qwen3.7 Plus (58.8%). The top models are clustered within 0.3 points, suggesting this benchmark is nearing saturation for frontier models.
7 models have been evaluated on NOVA-63. The benchmark falls in the Multilingual category. BenchAlign v6.0 gives NOVA-63 7% of the Multilingual reference weight, so it moves the Multilingual leaderboard. Reference weights are relative weights in the calibrated model, not fixed shares of a score.
About NOVA-63
Year
2026
Tasks
Broad multilingual evaluation
Format
Cross-lingual benchmark
Difficulty
Broad multilingual capability
NOVA-63 appears in multilingual comparison tables as a harder broad-language benchmark than standard translated math.
Freshness and provenance
Version
NOVA-63 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 NOVA-63 measure?
A broad multilingual benchmark row from Qwen's launch comparisons intended to measure cross-lingual capability beyond a single language family.
Which model scores highest on NOVA-63?
Qwen3.5 397B by Alibaba currently leads with a score of 59.1% on NOVA-63.
How many models are evaluated on NOVA-63?
7 AI models have published results on NOVA-63 in the BenchLM catalog.
Compare top models on NOVA-63
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.