JT-35B-Flash — benchmark results
China Mobile's flash-tier 35B dense model from the Jiutian family, positioned for cost-efficient high-throughput inference. Provider: China Mobile. Released 2026-05-14. Access: API.
Unified ELO 1728 ± 77, rank #225 of 1776 rated models, from 33 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA TAU-2 Bench | 99.12 | Accuracy (%) | 99.9 |
| AA Omniscience - Software Engineering (SWE) - Go | 36 | Accuracy (%) | 80.6 |
| AA Omniscience - Software Engineering (SWE) - Java | 29 | Accuracy (%) | 78.8 |
| AA GPQA Diamond | 82.93 | Accuracy (%) | 78.6 |
| AA Omniscience - Software Engineering (SWE) - HTML | 48 | Accuracy (%) | 77.7 |
| AA Omniscience - Software Engineering (SWE) - JavaScript | 41.82 | Accuracy (%) | 73.7 |
| AA Omniscience - Software Engineering (SWE) - C | 53 | Accuracy (%) | 73 |
| Artificial Analysis Intelligence Index | 28.36 | Intelligence Index | 72.9 |
| AA Omniscience - Software Engineering (SWE) - PHP | 38 | Accuracy (%) | 72.8 |
| AA Omniscience - Software Engineering (SWE) | 35 | Accuracy (%) | 71.9 |
| AA Omniscience - Software Engineering (SWE) - Kotlin | 28 | Accuracy (%) | 71.9 |
| AA Omniscience - Software Engineering (SWE) - TypeScript | 36.67 | Accuracy (%) | 71.8 |
Interactive version: theaggregate.ai/model?slug=jt-35b-flash · How the rankings work · Data refreshed daily, snapshot 2026-07-22.