LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

Gemma 4 31B vs GPT-5 Nano

Updated October 21, 2026. 48 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace Gemma 4 31B Replace GPT-5 Nano
Gemma 4 31BGoogle
Coding
38.3
Overall
14.7
Output $/M
$0.40
45
GPT-5 NanoOpenAI
Coding
34.8
Overall
13.0
Output $/M
$0.40
2
vs boards won of 48 shared, 1 tied

Decision reading

Pick Gemma 4 31B for coding: it scores 38.3 on the Vibe Coding Index against 34.8 for GPT-5 Nano.

Pick GPT-5 Nano when the context window is larger, 400K against 262K tokens, or when what matters is OTIS Mock AIME or SimpleQA Verified, where it leads.

Model AGemma 4 31BGoogle38.3/1002 independent benchmarksNo cost per solved taskModel BGPT-5 NanoOpenAI34.8/1001 independent benchmark$0.109 per solved task
Shared results
48
Gemma 4 31B only
20
GPT-5 Nano only
37
Like-for-like categories
7 of 9

Shared results can support a head-to-head reading. Results only one model has describe coverage, not superiority.

Which one for your work

Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.

At a glance

Coding index
Gemma 4 31B 38.3 from 2 benchmarks, GPT-5 Nano 34.8 from 1 benchmark.
Widest category
Agentic work, 28.9 points apart on average over 1 shared benchmark.
Widest single gap
LMCA: Gemma 4 31B 39.3% against GPT-5 Nano 7.9%, 31.3 points apart on a 0 to 100 scale.
Output price
Both cost $0.40 per million output tokens.
Context window
Gemma 4 31B takes 262K tokens, GPT-5 Nano 400K.

Benchmark by benchmark

First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.

Boards won, where both have a result
CategoryGemma 4 31BBarsGPT-5 NanoTies
Coding40
Intelligence20
Agentic work10
Reasoning80
Math21
Knowledge511 tie
General230
All shared boards4521 tie
CodingLike-for-likeFour of these boards feed the Vibe Coding Index40
Coding benchmarks for Gemma 4 31B and GPT-5 Nano
BenchmarkGemma 4 31BGPT-5 NanoLeader
SWE-bench VerifiedNot measured34.8%Only GPT-5 Nano measured
LiveBench coding49.0%Not measuredOnly Gemma 4 31B measured
LMArena WebDev1367Not measuredOnly Gemma 4 31B measured
AA Coding Index43.4Not measuredOnly Gemma 4 31B measured
LiveCodeBench (Vals AI)Not measured70.2%Only GPT-5 Nano measured
LiveCodeBench (Artificial Analysis run)Not measured78.9%Only GPT-5 Nano measured
SciCode45.5%Not measuredOnly Gemma 4 31B measured
Terminal-Bench Hard (Artificial Analysis run)36.4%17.4%Gemma 4 31B
Terminal-BenchNot measured21.8%Only GPT-5 Nano measured
WeirdML52.3%38.1%Gemma 4 31B
ALE-Bench925.5Not measuredOnly Gemma 4 31B measured
LMArena Text: coding14591351Gemma 4 31B
LMArena Text: software and IT14631345Gemma 4 31B
LMArena WebDev: React1359Not measuredOnly Gemma 4 31B measured
LMArena WebDev: HTML1380Not measuredOnly Gemma 4 31B measured
Neither model has a result on the other 23 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
Intelligence benchmarks for Gemma 4 31B and GPT-5 Nano
BenchmarkGemma 4 31BGPT-5 NanoLeader
AA Intelligence Index14.713.0Gemma 4 31B
Epoch Capabilities Index143.0139.2Gemma 4 31B
Agentic workLike-for-likePublic boards, shown as published, kept out of the index10
Agentic work benchmarks for Gemma 4 31B and GPT-5 Nano
BenchmarkGemma 4 31BGPT-5 NanoLeader
AA Agentic Index4.2Not measuredOnly Gemma 4 31B measured
LiveBench research44.4%Not measuredOnly Gemma 4 31B measured
LiveBench browser4.0%Not measuredOnly Gemma 4 31B measured
LiveBench long agentic13.3%Not measuredOnly Gemma 4 31B measured
τ²-Bench Telecom65.5%36.5%Gemma 4 31B
Neither model has a result on the other 34 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index80
Reasoning benchmarks for Gemma 4 31B and GPT-5 Nano
BenchmarkGemma 4 31BGPT-5 NanoLeader
GPQA DiamondNot measured63.4%Only GPT-5 Nano measured
LiveBench reasoning68.3%Not measuredOnly Gemma 4 31B measured
MMMU Pro (Vals AI)Not measured70.9%Only GPT-5 Nano measured
GPQA Diamond (Artificial Analysis run)85.7%67.6%Gemma 4 31B
Humanity's Last Exam (Artificial Analysis run)23.6%9.5%Gemma 4 31B
GPQA Diamond (Epoch AI run)75.8%69.4%Gemma 4 31B
ARC-AGINot measured20.7%Only GPT-5 Nano measured
ARC-AGI-2Not measured2.6%Only GPT-5 Nano measured
Chess puzzlesNot measured22.0%Only GPT-5 Nano measured
Mystery game puzzlesNot measured9.0%Only GPT-5 Nano measured
VPCTNot measured37.2%Only GPT-5 Nano measured
LMArena Text: hard prompts14481328Gemma 4 31B
LMArena Text: expert14641321Gemma 4 31B
LMArena Text: hard prompts in English14561338Gemma 4 31B
LMArena Vision: diagrams12831141Gemma 4 31B
LMArena Vision: homework13101157Gemma 4 31B
Neither model has a result on the other 5 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index21
Math benchmarks for Gemma 4 31B and GPT-5 Nano
BenchmarkGemma 4 31BGPT-5 NanoLeader
AIMENot measured81.2%Only GPT-5 Nano measured
MATH 500 (Vals AI)Not measured93.8%Only GPT-5 Nano measured
MGSM (Vals AI)Not measured89.3%Only GPT-5 Nano measured
AA Math IndexNot measured83.7Only GPT-5 Nano measured
AIME 2025 (Artificial Analysis run)Not measured83.7%Only GPT-5 Nano measured
OTIS Mock AIME73.3%81.1%GPT-5 Nano
FrontierMath, tiers 1 to 3Not measured20.0%Only GPT-5 Nano measured
FrontierMath, tier 4Not measured2.4%Only GPT-5 Nano measured
ProofBenchNot measured12.0%Only GPT-5 Nano measured
MATH level 5Not measured95.2%Only GPT-5 Nano measured
LMArena Text: math14631317Gemma 4 31B
LMArena Text: mathematical work14651326Gemma 4 31B
Neither model has a result on the other 1 board here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index51
Knowledge benchmarks for Gemma 4 31B and GPT-5 Nano
BenchmarkGemma 4 31BGPT-5 NanoLeader
MMLU ProNot measured76.1%Only GPT-5 Nano measured
LegalBench (Vals AI)Not measured50.1%Only GPT-5 Nano measured
CaseLaw (Vals AI)52.6%52.6%Tie
TaxEval (Vals AI)Not measured67.4%Only GPT-5 Nano measured
MortgageTax (Vals AI)61.4%53.6%Gemma 4 31B
MedQA (Vals AI)Not measured93.3%Only GPT-5 Nano measured
MedCode (Vals AI)Not measured30.4%Only GPT-5 Nano measured
MMLU Pro (Artificial Analysis run)Not measured78.0%Only GPT-5 Nano measured
SimpleQA Verified10.4%11.7%GPT-5 Nano
LMArena Text: science14571318Gemma 4 31B
LMArena Text: medicine and healthcare14511335Gemma 4 31B
LMArena Text: legal and government14381298Gemma 4 31B
LMArena Text: business and finance14301311Gemma 4 31B
Neither model has a result on the other 4 boards here
Web designDirectional onlyPublic boards, shown as published, kept out of the index7 boards, none shared
Web design benchmarks for Gemma 4 31B and GPT-5 Nano
BenchmarkGemma 4 31BGPT-5 NanoLeader
Design Arena: websitesNot measured1101Only GPT-5 Nano measured
Design Arena: UI componentsNot measured1066Only GPT-5 Nano measured
Design Arena: game devNot measured1045Only GPT-5 Nano measured
Design Arena: data vizNot measured1056Only GPT-5 Nano measured
Design Arena: 3DNot measured966Only GPT-5 Nano measured
Design Arena: all code categoriesNot measured1084Only GPT-5 Nano measured
LiveBench design58.4%Not measuredOnly Gemma 4 31B measured
Neither model has a result on the other 2 boards here
GeneralLike-for-likePublic boards, shown as published, kept out of the index230
General benchmarks for Gemma 4 31B and GPT-5 Nano
BenchmarkGemma 4 31BGPT-5 NanoLeader
LiveBench data analysis53.3%Not measuredOnly Gemma 4 31B measured
SAGE (Vals AI)55.0%30.4%Gemma 4 31B
MedScribe (Vals AI)Not measured72.9%Only GPT-5 Nano measured
IFBench75.6%67.6%Gemma 4 31B
Long context reasoning (AA-LCR)69.7%45.0%Gemma 4 31B
DTBench82.7%62.7%Gemma 4 31B
LMCA39.3%7.9%Gemma 4 31B
Fiction.LiveBenchNot measured44.4%Only GPT-5 Nano measured
Surface Evolver Bench30.6%Not measuredOnly Gemma 4 31B measured
LMArena Text14431320Gemma 4 31B
LMArena Text: instruction following14331306Gemma 4 31B
LMArena Text: creative writing14171249Gemma 4 31B
LMArena Vision12761159Gemma 4 31B
LMArena Text: English14511321Gemma 4 31B
LMArena Text: non-English14301313Gemma 4 31B
LMArena Text: Chinese14751356Gemma 4 31B
LMArena Text: GermanNot measured1327Only GPT-5 Nano measured
LMArena Text: JapaneseNot measured1227Only GPT-5 Nano measured
LMArena Text: KoreanNot measured1269Only GPT-5 Nano measured
LMArena Text: French1436Not measuredOnly Gemma 4 31B measured
LMArena Text: Spanish14421360Gemma 4 31B
LMArena Text: Russian14601296Gemma 4 31B
LMArena Text: Polish14331314Gemma 4 31B
LMArena Text: multi-turn14521311Gemma 4 31B
LMArena Text: longer queries14461312Gemma 4 31B
LMArena Text: writing and language14321283Gemma 4 31B
LMArena Text: entertainment and media14051268Gemma 4 31B
LMArena Vision: English12711170Gemma 4 31B
LMArena Vision: Chinese13261176Gemma 4 31B
LMArena Vision: reading text12821152Gemma 4 31B
LMArena Vision: humor12801110Gemma 4 31B
LMArena Vision: captioning1218Not measuredOnly Gemma 4 31B measured
LMArena Vision: recognizing things1226Not measuredOnly Gemma 4 31B measured
LMArena Vision: creative writing1285Not measuredOnly Gemma 4 31B measured
LMArena Document1425Not measuredOnly Gemma 4 31B measured
Neither model has a result on the other 11 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for Gemma 4 31B and GPT-5 Nano
BenchmarkGemma 4 31BGPT-5 NanoLeader
Neither model has a result on 5 boards here

What each workload costs

Three fixed token mixes at each model's listed price, so per-token rates become a bill.

Chat turn

1K input + 500 output tokens

Gemma 4 31B$0.0004Fits in one request
GPT-5 Nano$0.0003Fits in one request

GPT-5 Nano costs less

Repository review

50K input + 3K output tokens

Gemma 4 31B$0.011Fits in one request
GPT-5 Nano$0.0037Fits in one request

GPT-5 Nano costs less

Agent loop

200K input + 10K output tokens

Gemma 4 31B$0.044Fits in one request
GPT-5 Nano$0.014Fits in one request

GPT-5 Nano costs less

Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.

Specification differences

The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.

Specifications of Gemma 4 31B and GPT-5 Nano
SpecificationGoogle Gemma 4 31BOpenAI GPT-5 Nano
MakerGoogleOpenAI
WeightsOpen weightsNo public weights listed
ReleasedApril 2, 2026August 7, 2025
Context window262K400K
Input $/Mtok$0.20$0.05
Output $/Mtok$0.40$0.40
Coding index38.334.8
Independent measurements behind it21
Cost per solved taskNot computable$0.109
SourceOpenRouter listingOpenRouter listing

Questions

Is Gemma 4 31B better than GPT-5 Nano for coding?

Pick Gemma 4 31B for coding: it scores 38.3 on the Vibe Coding Index against 34.8 for GPT-5 Nano. Pick GPT-5 Nano when the context window is larger, 400K against 262K tokens, or when what matters is OTIS Mock AIME or SimpleQA Verified, where it leads.

Which is cheaper, Gemma 4 31B or GPT-5 Nano?

Gemma 4 31B costs $0.20 per million input tokens and $0.40 per million output tokens; GPT-5 Nano costs $0.05 and $0.40. Cost per solved task is known for GPT-5 Nano only.

Which has the bigger context window, Gemma 4 31B or GPT-5 Nano?

GPT-5 Nano, with 400K tokens against 262K.

How many benchmarks is the Gemma 4 31B vs GPT-5 Nano comparison based on?

48 benchmarks have a result for both models as of October 21, 2026. Gemma 4 31B's coding index rests on 2 independent benchmarks and GPT-5 Nano's on 1. Where only one model has a result, the page says so instead of naming a winner.

Compare next