Ling 3.1 Flash: benchmark results
InclusionAI's Ling 3.1 Flash (September 2026), a 560B-total, 25B-active hybrid-reasoning MoE served by API (free trial through October 13); InclusionAI has said it will open-source the weights after the trial, but none are published yet. Provider: InclusionAI. Released 2026-09-30. Access: API.
Aggregate Rating 1788 ± 32, rank #72 of 1635 rated models, from 72 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenRouter Tau2-Bench Airline | 79.3 | Accuracy (%) | 98.6 |
| AI Chess Leaderboard (Reasoning) | 1800 | Elo | 98.5 |
| AutomationBench-AA - Marketing - Raw Objective Completion | 89.72 | Raw Objective Completion (%) | 97.3 |
| AutomationBench-AA - Finance - Raw Objective Completion | 90.3 | Raw Objective Completion (%) | 96.4 |
| AA Long Context Reasoning | 83 | Accuracy (%) | 95 |
| AutomationBench-AA - HR - Raw Objective Completion | 76.18 | Raw Objective Completion (%) | 95 |
| Artificial Analysis Intelligence Index | 41.09 | Intelligence Index | 92 |
| AA GDPval | 1617.9 | ELO | 91.4 |
| The Aggregate - Puzzle Portfolio Ladder | 96.3 | Accuracy (%) | 88.9 |
| AutomationBench-AA | 61.74 | Guardrail-Adjusted Objective Completion (%) | 88.6 |
| AA-Omniscience Hallucination Rate | 37.88 | Hallucination Rate (%) | 88.1 |
| AA-Omniscience Index - Software Engineering (SWE) - Swift | 56 | Omniscience Index | 88.1 |
Interactive version: theaggregate.ai/model?slug=ling-3-1-flash · How It Works · Data refreshed daily, snapshot 2026-10-15.