Ling-2.6-flash: benchmark results
InclusionAI's open 104B MoE (7.4B active) with hybrid linear+MLA attention, optimized for inference efficiency and agent workloads (April 2026). Provider: InclusionAI. Released 2026-04-22. Access: Open.
Aggregate Rating 1573 ± 21, rank #532 of 1656 rated models, from 90 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| FinRiskAtlas - Applied Review - Risk Classification | 75 | Accuracy (%) | 81.2 |
| FinRiskAtlas - FinRisk-Ask - Balanced Action Agreement | 56.03 | BAcc (%) | 81.2 |
| AA TAU-2 Bench | 85.96 | Accuracy (%) | 78.2 |
| Tau2-Bench Telecom | 85.96 | Success Rate (%) | 78.1 |
| AA IFBench | 57.41 | Accuracy (%) | 68.8 |
| AA Omniscience - Software Engineering (SWE) - R | 14 | Accuracy (%) | 67.6 |
| AA Omniscience - Software Engineering (SWE) - Java | 18 | Accuracy (%) | 66.9 |
| AA Terminal-Bench Hard | 21.21 | Accuracy (%) | 60.9 |
| AA Omniscience - Software Engineering (SWE) - Python | 20.5 | Accuracy (%) | 59.7 |
| AA Omniscience - Software Engineering (SWE) - Rust | 50 | Accuracy (%) | 59 |
| AA Omniscience - Software Engineering (SWE) - JavaScript | 29.09 | Accuracy (%) | 57 |
| AA Omniscience - Software Engineering (SWE) - C | 34 | Accuracy (%) | 56.5 |
Interactive version: theaggregate.ai/model?slug=ling-2-6-flash · How It Works · Data refreshed daily, snapshot 2026-10-20.