Sage-S2.1: benchmark results

Provider: Shanghai AI Lab. Access: Open.

Aggregate Rating 1701 ± 29, rank #325 of 10489 rated models, from 96 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Nejumi 4 - MT-Bench (Japanese) - Extraction98Judge rating (1-10, x10)91.2
Nejumi 4 - ALT - Bias93.44Score (%)89.4
Nejumi 4 - jaster (0-shot) - JCoLA (out-of-domain)89Exact match (%)87.1
Nejumi 4 - GLP - Function Calling67.96Score (%)86.1
Nejumi 4 - BFCL - Live AST73.15Accuracy (%)83.6
Nejumi 4 - GLP - Application Development66.41Score (%)80.8
Nejumi 4 - jaster (0-shot) - ALT Japanese-to-English90COMET wmt22 (x100)80.2
Nejumi 4 - jaster (0-shot) - JCommonsenseQA99Exact match (%)80.2
Nejumi 4 - Toxicity - Fairness98.31Criteria met (%)79.9
Nejumi 4 - jaster (2-shot) - JCommonsenseQA99Exact match (%)77.7
Nejumi 4 - jaster (2-shot) - AIO92.9Character F1 (x100)74.2
Nejumi 4 - BFCL - Multi-Turn52.22Accuracy (%)73.9

Interactive version: theaggregate.ai/model?slug=sage-s2-1 · How It Works · Data refreshed daily, snapshot 2026-11-20.