LLMs TigerLive LLM benchmarks

Gemma 4 31B IT: benchmark scores, price and ranking

Gemma 4 31B IT from Google DeepMind ranks #443 of 743 LLMs on the LLMs Tiger aggregate benchmark score (47.3), based on 13 benchmark results captured up to 2026-10-13.

Organisation
Google DeepMind
Type
Open-weight / local model
Open weights
Yes
Released
2026-04-02 (source: epoch)
Cheapest API price
$0.100 in / $0.340 out per million tokens (via wandb, observed 2026-10-13T19:47Z)
Aggregate rank
#443 of 743 (score 47.3)

Relative to other models, Gemma 4 31B IT is strongest on Epoch ECI (#38 of 204) and weakest on SimpleQA Verified (#67 of 69).

Across the 13 benchmarks where at least ten models are scored, Gemma 4 31B IT is in the top 10% on 0 and in the top half on 7.

Its cheapest listed input price, $0.100 per million tokens, is lower than that of 79% of the 427 models with a listed price.

Gemma 4 31B IT benchmark results

Gemma 4 31B IT benchmark scores, ranked against all models measured on each benchmark
BenchmarkScoreRankSourceCaptured
Epoch ECI143#38 of 204Epoch AI Benchmarking Hub2026-10-13
WeirdML52.3%#28 of 109Epoch AI Benchmarking Hub2026-10-13
LMCA39.2%#31 of 108Epoch AI Benchmarking Hub2026-10-13
DTBench82.7%#42 of 146Epoch AI Benchmarking Hub2026-10-13
ALE-Bench926#27 of 79Epoch AI Benchmarking Hub2026-10-13
OTIS Mock AIME73.3%#71 of 189Epoch AI Benchmarking Hub2026-10-13
GPQA Diamond75.8%#124 of 275Epoch AI Benchmarking Hub2026-10-13
SciCode43.4%#55 of 106Epoch AI Benchmarking Hub2026-10-13
CritPt1.4%#63 of 115Epoch AI Benchmarking Hub2026-10-13
WebDev Arena1,364#43 of 67Epoch AI Benchmarking Hub2026-10-13
Chess Puzzles5.0%#79 of 122Epoch AI Benchmarking Hub2026-10-13
Surface Evolver30.6%#10 of 13Epoch AI Benchmarking Hub2026-10-13
SimpleQA Verified10.4%#67 of 69Epoch AI Benchmarking Hub2026-10-13

About Gemma 4 31B IT

Gemma 4 31B IT is an open-weight model that can run on your own hardware from Google DeepMind. It was released on 2026-04-02, according to Epoch AI. The cheapest API listing we track charges $0.100 per million input tokens and $0.340 per million output tokens, through wandb. We hold 13 benchmark results for it across 4 areas, and none is in the top tenth of its board.

Where Gemma 4 31B IT stands

On our aggregate score, Gemma 4 31B IT is #443 of 743 ranked models, around the middle of the table. Its score is 47.3; the scale is centred on 50, so it sits 2.7 points below the typical model. That beats 40% of the ranked field. Among the 41 ranked Google DeepMind models it is #22. The next one up is Gemma 4 26B A4B at 47.5. The next one down is Gemini 2.0 Pro Exp (Feb 2025) at 46.7.

Gemma 4 31B IT, benchmark by benchmark

Epoch ECI puts Gemma 4 31B IT at 143, #38 of 204. That beats 81% of the models tested, but 14 points short of Qwen 3.8 Max (156). The results either side are Gemini 2.5 Flash (Sep 2025) (143) and Claude Opus 4 (143). It covers Epoch Capabilities Index: Epoch's own aggregate of ~40 benchmarks. The board median is 127, so Gemma 4 31B IT is 15 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

On WeirdML, Gemma 4 31B IT reaches 52.3%. It ranks #28 of 109, ahead of 74% of the field, and 41.0 points behind GPT-6 Astra (max) at 93.3%. The results either side are o3 (high) (52.4%) and grok-4-20 (52.3%). It covers unusual machine-learning tasks. The board median is 41.2%, so Gemma 4 31B IT is 11.1 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

Gemma 4 31B IT manages 39.2% on LMCA, good for #31 of 108. Claude Opus 5.5 (max) leads at 68.2%, 29.0 points further up, and 71% of models sit below. The results either side are o3 (high) (39.7%) and grok-4-20 (38.7%). It covers language-model capability assessment. The board median is 29.2%, so Gemma 4 31B IT is 10.0 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

DTBench puts Gemma 4 31B IT at 82.7%, #42 of 146. That beats 71% of the models tested, but 16.3 points short of Claude Opus 5.5 (max) (98.9%). The results either side are Qwen3.5 Flash (82.9%) and grok-4-fast (82.7%). It covers dynamic tool-use benchmark. The board median is 68.8%, so Gemma 4 31B IT is 13.9 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

On ALE-Bench, Gemma 4 31B IT reaches 926. It ranks #27 of 79, ahead of 66% of the field, and 2,026 points behind GPT-6 Astra (max) at 2,951. The results either side are Gemma 4 26B A4B (927) and Gemini 3.7 Flash (high) (904). It covers AtCoder heuristic-optimisation contests (performance rating). The board median is 764, so Gemma 4 31B IT is 162 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

Gemma 4 31B IT manages 73.3% on OTIS Mock AIME, good for #71 of 189. Claude Fable 5.1 (max) leads at 100.0%, 26.7 points further up, and 62% of models sit below. The results either side are o3-mini (high) (76.9%) and o1 (high) (73.3%). It covers competition maths (AIME 2024-25 style). The board median is 57.8%, so Gemma 4 31B IT is 15.6 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

Gemma 4 31B IT lands below the middle on WebDev Arena: 1,364, #43 of 67, with 63% of the field scoring higher. Claude Opus 5.5 (max) leads at 1,820; the closest result above is grok-4-20 at 1,374. It covers building web apps, crowd-voted (Elo). The board median is 1,433, so Gemma 4 31B IT is 69 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

Gemma 4 31B IT lands below the middle on Chess Puzzles: 5.0%, #79 of 122, with 64% of the field scoring higher. GPT-6 Astra (max) leads at 72.0%; the closest result above is claude-3-opus-20240229 at 5.0%. It covers chess tactics puzzles. The board median is 12.5%, so Gemma 4 31B IT is 7.5 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

Gemma 4 31B IT lands below the middle on Surface Evolver: 30.6%, #10 of 13, with 69% of the field scoring higher. Kimi K3 (max) leads at 93.0%; the closest result above is Qwen3.6 35B-A3B at 44.4%. It covers geometry optimisation with a solver. The board median is 46.2%, so Gemma 4 31B IT is 15.6 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

SimpleQA Verified is a weak spot. Gemma 4 31B IT scores 10.4%, #67 of 69, behind 96% of the models measured. The leader, gemini-4-argon-high, reaches 76.0%. It covers short factual questions. The board median is 41.0%, so Gemma 4 31B IT is 30.6 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

The full table above also lists the other 3 results, including boards with fewer than ten models, where a rank says little.

Strengths and weaknesses

Software engineering: uneven, ahead of 50% of the field on average across 3 benchmarks. Its best board here is ALE-Bench (#27 of 79) and its weakest is WebDev Arena (#43 of 67). Reasoning, maths and science: solid, ahead of 54% of the field on average across 3 benchmarks. Its best board here is OTIS Mock AIME (#71 of 189) and its weakest is CritPt (#63 of 115). General knowledge and chat: uneven, ahead of 48% of the field on average across 6 benchmarks. Its best board here is Epoch ECI (#38 of 204) and its weakest is SimpleQA Verified (#67 of 69).

Price and value

The median input price across the 427 priced models is $0.500 per million tokens, so Gemma 4 31B IT costs less than the typical listing. Models that score within 1.5 points or better and cost less: Qwen2.5-Coder-7B-Instruct ($0.010, 49.5), DeepSeek V4.1 Flash (high) ($0.010, 48.9), Qwen2.5-Coder-3B-Instruct ($0.010, 46.5). Similar scores cost more on GLM-4.5-Air ($0.125) and Qwen2-VL-72B-Instruct ($0.130).

Release history

The Google DeepMind model released before it was Gemma 4 26B A4B on 2026-04-02, which scores 47.5. The next one was AI co-mathematician on 2026-05-08, at 52.3.

Reading these numbers

The 13 scores on this page come from 1 source: Epoch AI Benchmarking Hub (13). No official leaderboard supplied any of them, so treat them as third-party measurements.

Models ranked nearby

More from Google DeepMind

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