Benchmark Almanac

Gemini 3.8 Live

Gemini 3.8 Live ranks first on Jarvis Bench.

Google DeepMind Released September 15th, 2026 Family Gemini 3.8 Audio The lab's model card Compare with others
11
Benchmarks with a score
1
Ranked first, out of five or more models
1
In the top three
Google DeepMind Sep 2026

Gemini 3.8 Live

Closed weights Mid-size

No Epoch Capabilities Index score yet
  • Realtime-Bench 6.00% #11 /11
  • Cekura Turn Bench 81.5% #5 /8
  • Cekura Agent Performance 41.5% #7 /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.

Agents and tools

BenchmarkScoreRankSource
VAmoS Pro Bench 40.1% #1 of 3 Official leaderboard
Speko Speech-to-Speech Benchmark (six concierge scenarios) 0.77 #9 of 11 Official leaderboard
Realtime-Bench 6.00% #11 of 11 Official leaderboard
MIVAS Bench 43.5% #4 of 4 Official leaderboard

Vision, audio and robotics

BenchmarkScoreRankSource
Jarvis Bench 998 #1 of 5 Official leaderboard
Voice Arena Full Duplex Bench 99.9% #1 of 2 Official leaderboard
Cekura Real Human Calls 24.3% #4 of 7 Official leaderboard
Cekura Turn Bench 81.5% #5 of 8 Official leaderboard
EVA-Bench 0.46 #7 of 11 Official leaderboard
VoxParity 0.07 #9 of 14 Official leaderboard
Cekura Agent Performance 41.5% #7 of 7 Official leaderboard

Who measured it?

Independent runs cover all 11 of its benchmarks.

  • An independent run only 11
Official leaderboards scored it on 11.

Sources: Official leaderboards, Epoch AI (CC BY 4.0) and labs’ model cards