B.O.B. Leaderboard

Google DeepMindreleased 2 Apr 2026

Gemma 4 26B A4B

#87 of 9514 of 46 benchmarksConsistent

11.0

B.O.B.

Category profile

  • Real use (hard to game)2×11
  • Reasoning & knowledge2
  • Math13
  • Coding21
  • Agents & real work7
  • Independent indexes12

Reality check

Lab tests

11

Real use (3 tests)

11

Lab tests and real-use tests agree. Its lab-test and real-use scores are within 2 points of each other.

What counts as real use

Price per 1M tokens (OpenRouter)

$0.076 in · $0.26 out

Cost per task on the AA index

—

Output speed (Artificial Analysis)

68 tok/s

§Receipts

Every benchmark result

“Beats” = the share of tracked models this one beat on that test. The raw score is its best reported setting.

BenchmarkRawBeatsSource
Real use (hard to game) · Real use 11
LMArena Textas gemma-4-26b-a4b1437
20%of 80
LMArena (blind human votes)
LMArena WebDevas gemma-4-26b-a4b1359
14%of 82
LMArena (blind human votes)
EQ-Bench Creative Writingas google/gemma-4-26B-A4B-it1305
0%of 46
Sam Paech (open source)
Reasoning & knowledge · Reasoning 2
CritPt (physics research)as gemma-4-26b-a4b0.0%
4%of 89
Epoch AI Benchmarking Hub
Chess Puzzlesas gemma-4-26b-a4b-it_minimal6.0%
3%of 73
Epoch AI Benchmarking Hub
GPQA Diamondas gemma-4-26b-a4b-it_minimal73.2%
0%of 70
Epoch AI Benchmarking Hub
Math · Math 13
OTIS Mock AIMEas gemma-4-26b-a4b-it_minimal82.2%
13%of 73
Epoch AI Benchmarking Hub
Coding · Coding 21
ALE-Benchas gemma-4-26b-a4b927.2%
41%of 71
Epoch AI Benchmarking Hub
SciCodeas gemma-4-26b-a4b40.0%
16%of 86
Epoch AI Benchmarking Hub
WeirdMLas gemma-4-26b-a4b35.2%
6%of 52
Epoch AI Benchmarking Hub
Agents & real work · Agents 7
LMCAas gemma-4-26b-a4b29.7%
9%of 67
Epoch AI Benchmarking Hub
DTBenchas gemma-4-26b-a4b74.9%
5%of 67
Epoch AI Benchmarking Hub
Independent indexes · Indexes 12
Artificial Analysis Intelligence Indexas Gemma 4 26B A4B17
21%of 54
Artificial Analysis
Epoch Capabilities Indexas Gemma 4 26B A4B141.8
3%of 81
Epoch AI