Rankings / Google DeepMind
Gemma 4 31B
open weightsreleased 2026-04-02
BenchAtlas Index
as of 2026-09-11
thinking
rank #239
7 families · 4 categories · medium
no reasoning
rank #284
7 families · 4 categories · medium
Benchmark evidence
35 results
agentic coding
| τ²-Bench | no reasoning | 65.5 % | independent | Artificial Analysis |
| τ²-Bench | thinking | 59.9 % | independent | Artificial Analysis |
| τ²-Bench subset=banking | thinking | 14.8 % | independent | Artificial Analysis |
| τ²-Bench subset=banking | no reasoning | 8.9 % | independent | Artificial Analysis |
coding
| SciCode | thinking | 45.5 % | independent | Artificial Analysis |
| SciCode | no reasoning | 41.1 % | independent | Artificial Analysis |
external indices
| AA Coding Index | thinking | 43.4 points | independent | Artificial Analysis |
| AA Coding Index | no reasoning | 33.2 points | independent | Artificial Analysis |
| Intelligence Index v4.1 | thinking | 15.4 points | independent | Artificial Analysis |
| Intelligence Index v4.1 | no reasoning | 13.9 points | independent | Artificial Analysis |
human preference
| Coding (style control)older version arena=text · category=coding · style_control=true | 1498.9 1483.5–1514.2 | community | LMArena Leaderboard Dataset | |
| Industry Software And It Services (style control)older version arena=text · category=industry_software_and_it_services · style_control=true | 1491.4 1479.2–1503.6 | community | LMArena Leaderboard Dataset | |
| Hard Prompts English (style control)older version arena=text · category=hard_prompts_english · style_control=true | 1484.8 1468.8–1500.8 | community | LMArena Leaderboard Dataset | |
| Hard Prompts (style control)older version arena=text · category=hard_prompts · style_control=true | 1473.7 1463.8–1483.5 | community | LMArena Leaderboard Dataset | |
| Industry Life And Physical And Social Science (style control)older version arena=text · category=industry_life_and_physical_and_social_science · style_control=true | 1473.5 1454.0–1492.9 | community | LMArena Leaderboard Dataset | |
| Russian (style control)older version arena=text · category=russian · style_control=true | 1468.0 1447.6–1488.5 | community | LMArena Leaderboard Dataset | |
| Longer Query (style control)older version arena=text · category=longer_query · style_control=true | 1467.5 1453.8–1481.2 | community | LMArena Leaderboard Dataset | |
| Multi Turn (style control)older version arena=text · category=multi_turn · style_control=true | 1464.4 1446.6–1482.1 | community | LMArena Leaderboard Dataset | |
| English (style control)older version arena=text · category=english · style_control=true | 1462.6 1450.6–1474.6 | community | LMArena Leaderboard Dataset | |
| Exclude Ties (style control)older version arena=text · category=exclude_ties · style_control=true | 1454.3 1443.2–1465.4 | community | LMArena Leaderboard Dataset | |
| Instruction Following (style control)older version arena=text · category=instruction_following · style_control=true | 1452.4 1438.5–1466.3 | community | LMArena Leaderboard Dataset | |
| Overall (style control) arena=text · category=overall · style_control=true | 1451.3 1443.7–1458.9 | community | LMArena Leaderboard Dataset | |
| Industry Business And Management And Financial Operations (style control)older version arena=text · category=industry_business_and_management_and_financial_operations · style_control=true | 1445.8 1428.8–1462.7 | community | LMArena Leaderboard Dataset | |
| Non English (style control)older version arena=text · category=non_english · style_control=true | 1436.8 1427.3–1446.4 | community | LMArena Leaderboard Dataset | |
| Industry Writing And Literature And Language (style control)older version arena=text · category=industry_writing_and_literature_and_language · style_control=true | 1433.4 1417.6–1449.2 | community | LMArena Leaderboard Dataset | |
| Creative Writing (style control)older version arena=text · category=creative_writing · style_control=true | 1420.1 1400.8–1439.4 | community | LMArena Leaderboard Dataset | |
| Industry Entertainment And Sports And Media (style control)older version arena=text · category=industry_entertainment_and_sports_and_media · style_control=true | 1416.4 1398.4–1434.4 | community | LMArena Leaderboard Dataset |
knowledge science
| GPQA Diamond implementation=artificial-analysis | thinking | 85.7 % | independent | Artificial Analysis |
| GPQA Diamond implementation=artificial-analysis | no reasoning | 76.3 % | independent | Artificial Analysis |
| Humanity's Last Exam implementation=artificial-analysis | thinking | 23.6 % | independent | Artificial Analysis |
| Humanity's Last Exam implementation=artificial-analysis | no reasoning | 11.8 % | independent | Artificial Analysis |
long context instruction
| AA-LCR | thinking | 69.7 % | independent | Artificial Analysis |
| AA-LCR | no reasoning | 46.7 % | independent | Artificial Analysis |
| IFBench | thinking | 75.6 % | independent | Artificial Analysis |
| IFBench | no reasoning | 53.5 % | independent | Artificial Analysis |
Agent + model results
systems, not bare-model scores
| agent + model Artificial Analysis harness + Gemma 4 31B | Terminal-Bench 2.1 | 43.5 % | independent | Artificial Analysis |
| agent + model Artificial Analysis harness + Gemma 4 31B | Terminal-Bench Hard | 36.4 % | independent | Artificial Analysis |
| agent + model Artificial Analysis harness + Gemma 4 31B | Terminal-Bench Hard | 30.3 % | independent | Artificial Analysis |
| agent + model Artificial Analysis harness + Gemma 4 31B | Terminal-Bench 2.1 | 29.2 % | independent | Artificial Analysis |
These scores measure the whole agent system (scaffold, tools, budgets) — they are never merged into the bare model’s numbers.