Benchmark Almanac

Ling-3.1-flash

Ling-3.1-flash ranks first on 3 of the 32 benchmarks with five or more models scored.

Ant Group Released October 13th, 2026 Family Ling 3 The lab's model card Compare with others
33
Benchmarks with a score
3
Ranked first, out of five or more models
6
In the top three
Ant Group Oct 2026

Ling-3.1-flash

Open weights Mid-size

161.4 #24 of 404
Self-Reported Index, from the lab's own numbers. No Epoch Capabilities Index score yet.
  • WideSearch 83.3% #2 /19
  • HealthBench Professional 65.3% #6 /26
  • FinSearchComp 81.7% #1 /7

Where it ranks

Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.

Knowledge and language

BenchmarkScoreRankSource
WideSearch 83.3% #2 of 19 Lab model card
MultiChallenge 69.8% #5 of 42 Lab model card
Humanity's Last Exam 37.4% #68 of 154 Lab model card

Math and science

BenchmarkScoreRankSource
CritPt 16.3% #7 of 13 Lab model card
SuperMatBench 51.2% #1 of 1 Lab model card

Coding

BenchmarkScoreRankSource
DesignArena UI Component 1,351 #9 of 139 Official leaderboard
SWE-bench Multilingual 88.5% #6 of 73 Lab model card
FrontierSWE 75.2% #1 of 10 Lab model card
SWE-Bench Pro 68.7% #9 of 81 Lab model card
DesignArena 3D Design 1,324 #19 of 136 Official leaderboard
DesignArena Code 1,304 #20 of 143 Official leaderboard
DesignArena Website 1,293 #22 of 151 Official leaderboard
DesignArena Data Visualization 1,288 #18 of 118 Official leaderboard
DesignArena Game Dev 1,309 #25 of 143 Official leaderboard
DesignArena SVG 1,266 #21 of 103 Official leaderboard
SWE Atlas 64.5% #3 of 14 Lab model card
DesignArena Agentic Overall 1,198 #29 of 74 Official leaderboard
DesignArena React Native 1,192 #25 of 56 Official leaderboard
DAXBench v2 72.7% #32 of 47 Official leaderboard
DeepSWE 59.7% #40 of 56 Lab model card

Agents and tools

BenchmarkScoreRankSource
τ³-bench Banking 47.8% #1 of 22 Lab model card
BrowseComp 91.7% #5 of 81 Lab model card
SkillsBench 68.7% #2 of 14 Lab model card
AutomationBench 52.5% #9 of 47 Lab model card
MCP Atlas 83% #13 of 46 Lab model card
Terminal-Bench 4.0 46% #12 of 28 Lab model card
Terminal-Bench 2.1 81.2% #26 of 56 Lab model card
DRACO 85.5% #6 of 12 Lab model card

Professional work

BenchmarkScoreRankSource
FinSearchComp 81.7% #1 of 7 Lab model card
HealthBench Professional 65.3% #6 of 26 Lab model card
Vals Finance Agent v2 57.9% #5 of 11 Lab model card
GDPval-AA v2.1 1,673 #6 of 10 Lab model card

Safety and security

BenchmarkScoreRankSource
CyberGym 87.9% #5 of 34 Lab model card

Its lab's numbers put it at 161.4, number 24 of 404

The Self-Reported Index fits one score to the 19 capability results Ant Group reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 158.4 to 164.4. How we compute it.

Its launch post

The benchmarks the lab chose to headline, with the scores it gave.

Ling-3.1-flash model card (Hugging Face) Oct 13, 2026