LLMs TigerLive LLM benchmarks

Gemini 3 Pro Preview: benchmark scores, price and ranking

Gemini 3 Pro Preview from Google DeepMind ranks #30 of 741 LLMs on the LLMs Tiger aggregate benchmark score (53.9), based on 20 benchmark results captured up to 2026-10-05.

Organisation
Google DeepMind
Type
Preview release
Open weights
No
Cheapest API price
$2.00 in / $12.00 out per million tokens (via replicate, observed 2026-10-05T21:47Z)
Aggregate rank
#30 of 741 (score 53.9)

Relative to other models, Gemini 3 Pro Preview is strongest on SimpleBench (#2 of 64) and weakest on CL-bench (#11 of 16).

Gemini 3 Pro Preview benchmark results

Gemini 3 Pro Preview benchmark scores, ranked against all models measured on each benchmark
BenchmarkScoreRankSourceCaptured
SimpleBench76.4%#2 of 64Epoch AI Benchmarking Hub2026-10-05
Terminal-Bench 2.069.4%#1 of 28Epoch AI Benchmarking Hub2026-10-05
Humanity's Last Exam37.5%#2 of 29Epoch AI Benchmarking Hub2026-10-05
GPQA Diamond92.6%#19 of 274Epoch AI Benchmarking Hub2026-10-05
EnigmaEval18.2%#2 of 26Epoch AI Benchmarking Hub2026-10-05
METR time horizon224 min#3 of 34Epoch AI Benchmarking Hub2026-10-05
BALROG58.1%#4 of 34Epoch AI Benchmarking Hub2026-10-05
WeirdML69.9%#15 of 112Epoch AI Benchmarking Hub2026-10-05
SWE-bench Verified74.2%#10 of 73SWE-bench Verified leaderboard2026-10-05
GSO-Bench18.6%#3 of 19Epoch AI Benchmarking Hub2026-10-05
Vending-Bench 2$5,478#3 of 19Epoch AI Benchmarking Hub2026-10-05
OTIS Mock AIME91.4%#36 of 188Epoch AI Benchmarking Hub2026-10-05
Chess Puzzles31.0%#25 of 124Epoch AI Benchmarking Hub2026-10-05
ALE-Bench1,177#19 of 78Epoch AI Benchmarking Hub2026-10-05
CritPt6.9%#44 of 111Epoch AI Benchmarking Hub2026-10-05
WebDev Arena1,439#25 of 63Epoch AI Benchmarking Hub2026-10-05
Remote Labor Index1.2%#1 of 2Epoch AI Benchmarking Hub2026-10-05
ProofBench20.0%#22 of 43Epoch AI Benchmarking Hub2026-10-05
MCP Atlas70.3%#8 of 14Scale Labs MCP Atlas leaderboard2026-10-05
CL-bench15.8%#11 of 16Epoch AI Benchmarking Hub2026-10-05

Gemini 3 Pro Preview is also listed at 1 other reasoning-effort setting; this page shows the highest-effort row, which is the one the leaderboard keeps by default.

Models ranked nearby

Compare it on the live leaderboard → · How the aggregate score works