Enter opens the highlighted result. Esc closes.

Gemma 4 31B vs GPT-5.5

GPT-5.5 leads Gemma 4 31B on coding, 71.3 to 39.0 on the Vibe Coding Index, and Gemma 4 31B costs 75.0x less per output token ($0.40 vs $30.00 per million). Updated October 22, 2026; 69 benchmarks have a result for both.

Replace Gemma 4 31B Replace GPT-5.5
Gemma 4 31BGoogle
Coding
39.0
Overall
42.1
Output $/M
$0.40
1
GPT-5.5OpenAI
Coding
71.3
Overall
69.2
Output $/M
$30.00
68
vs boards won of 69 shared

Gemma 4 31B or GPT-5.5: which to pick

Pick GPT-5.5 for coding: it scores 71.3 on the Vibe Coding Index against 39.0 for Gemma 4 31B.

Pick Gemma 4 31B for output tokens 75.0x cheaper ($0.40 vs $30.00 per million) and work measured by SAGE (Vals AI), where it leads.

Model AGemma 4 31BGoogle39.0/1002 independent benchmarksNo cost per solved taskModel BGPT-5.5OpenAI71.3/1005 independent benchmarks$1.65 per solved task
Shared results
69
Gemma 4 31B only
2
GPT-5.5 only
65
Like-for-like categories
8 of 9

Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.

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 39.0 from 2 benchmarks, GPT-5.5 71.3 from 5 benchmarks.
Widest category
Agentic work, 30.6 points apart on average over 5 shared benchmarks.
Widest single gap
Surface Evolver Bench: Gemma 4 31B 30.6% against GPT-5.5 88.1%, 57.5 points apart on a 0 to 100 scale.
Output price
Gemma 4 31B $0.40 against GPT-5.5 $30.00 per million output tokens, 75.0x apart.
Context window
Gemma 4 31B takes 262K tokens, GPT-5.5 1.05M.

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-5.5Ties
Coding011
Intelligence02
Agentic work05
Reasoning010
Math03
Knowledge08
Web design01
General128
All shared boards168
CodingLike-for-likeSix of these boards feed the Vibe Coding Index011
Coding benchmarks for Gemma 4 31B and GPT-5.5
BenchmarkGemma 4 31BGPT-5.5Leader
LiveBench coding49.0%85.1%GPT-5.5
LMArena WebDev13671513GPT-5.5
AA Coding Index43.474.9GPT-5.5
LiveCodeBench (Vals AI)Not measured85.3%Only GPT-5.5 measured
SWE-bench Verified (Vals AI)Not measured82.6%Only GPT-5.5 measured
Vibe Code Bench (Vals AI)Not measured69.8%Only GPT-5.5 measured
Code migration (Vals AI)Not measured45.2%Only GPT-5.5 measured
ProgramBench (Vals AI)Not measured0.5%Only GPT-5.5 measured
Terminal-Bench 2.1 (Vals AI)Not measured76.4%Only GPT-5.5 measured
SkillsBench (Vals AI)Not measured62.5%Only GPT-5.5 measured
SciCode45.5%56.1%GPT-5.5
Terminal-Bench Hard (Artificial Analysis run)36.4%60.6%GPT-5.5
Terminal-BenchNot measured84.7%Only GPT-5.5 measured
FrontierCodeNot measured43.0%Only GPT-5.5 measured
GSO-BenchNot measured40.2%Only GPT-5.5 measured
WeirdML52.3%84.9%GPT-5.5
MirrorCodeNot measured10.0%Only GPT-5.5 measured
ALE-Bench925.51943.0GPT-5.5
LMArena Text: coding14591494GPT-5.5
LMArena Text: software and IT14631493GPT-5.5
LMArena WebDev: React13591506GPT-5.5
LMArena WebDev: HTML13801543GPT-5.5
LMArena WebDev: from an imageNot measured1532Only GPT-5.5 measured
SciCode8.9%Not measuredOnly Gemma 4 31B measured
Neither model has a result on the other 15 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
Intelligence benchmarks for Gemma 4 31B and GPT-5.5
BenchmarkGemma 4 31BGPT-5.5Leader
AA Intelligence Index14.738.4GPT-5.5
Epoch Capabilities Index143.0159.7GPT-5.5
Agentic workLike-for-likePublic boards, shown as published, kept out of the index05
Agentic work benchmarks for Gemma 4 31B and GPT-5.5
BenchmarkGemma 4 31BGPT-5.5Leader
AA Agentic Index4.236.4GPT-5.5
Design Arena: agents building web appsNot measured1117Only GPT-5.5 measured
Design Arena: agents building full-stack appsNot measured1089Only GPT-5.5 measured
Design Arena: agents building mobile appsNot measured1142Only GPT-5.5 measured
Design Arena: agents building Android appsNot measured1164Only GPT-5.5 measured
Design Arena: agents building Godot gamesNot measured1173Only GPT-5.5 measured
Design Arena: agents building gamesNot measured1179Only GPT-5.5 measured
Design Arena: HTML slidesNot measured1017Only GPT-5.5 measured
Design Arena: PowerPoint slidesNot measured1084Only GPT-5.5 measured
LiveBench research44.4%61.6%GPT-5.5
LiveBench browser4.0%48.0%GPT-5.5
LiveBench long agentic13.3%44.6%GPT-5.5
SREBench (Vals AI)Not measured3.8%Only GPT-5.5 measured
Finance Agent (Vals AI)Not measured51.8%Only GPT-5.5 measured
Tax Agent Bench (Vals AI)Not measured21.8%Only GPT-5.5 measured
Legal Research Bench (Vals AI)Not measured40.4%Only GPT-5.5 measured
Excel Modeling Benchmark (Vals AI)Not measured64.5%Only GPT-5.5 measured
Time Horizon Index (Vals AI)Not measured8.3%Only GPT-5.5 measured
τ²-Bench Telecom65.5%93.9%GPT-5.5
APEX-AgentsNot measured55.1%Only GPT-5.5 measured
Remote Labor IndexNot measured6.3%Only GPT-5.5 measured
OSWorld 2.0Not measured13.0%Only GPT-5.5 measured
DeepResearch BenchNot measured54.0%Only GPT-5.5 measured
PostTrainBenchNot measured27.2%Only GPT-5.5 measured
Vending-Bench 2Not measured$7,524Only GPT-5.5 measured
Neither model has a result on the other 16 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index010
Reasoning benchmarks for Gemma 4 31B and GPT-5.5
BenchmarkGemma 4 31BGPT-5.5Leader
GPQA DiamondNot measured93.2%Only GPT-5.5 measured
LiveBench reasoning68.3%89.7%GPT-5.5
MMMU Pro (Vals AI)Not measured88.3%Only GPT-5.5 measured
GPQA Diamond (Artificial Analysis run)85.7%93.5%GPT-5.5
Humanity's Last Exam (Artificial Analysis run)23.6%45.8%GPT-5.5
GPQA Diamond (Epoch AI run)75.8%90.7%GPT-5.5
ARC-AGINot measured95.0%Only GPT-5.5 measured
ARC-AGI-2Not measured85.0%Only GPT-5.5 measured
SimpleBenchNot measured69.0%Only GPT-5.5 measured
Chess puzzlesNot measured63.0%Only GPT-5.5 measured
Mystery game puzzlesNot measured56.0%Only GPT-5.5 measured
EBR-benchNot measured34.3%Only GPT-5.5 measured
LMArena Text: hard prompts14481489GPT-5.5
LMArena Text: expert14641507GPT-5.5
LMArena Text: hard prompts in English14561483GPT-5.5
LMArena Vision: diagrams12831325GPT-5.5
LMArena Vision: homework13101332GPT-5.5
Kagi LLM Benchmark63.5%88.8%GPT-5.5
Neither model has a result on the other 4 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index03
Math benchmarks for Gemma 4 31B and GPT-5.5
BenchmarkGemma 4 31BGPT-5.5Leader
OTIS Mock AIME73.3%84.4%GPT-5.5
FrontierMath, tiers 1 to 3Not measured85.3%Only GPT-5.5 measured
FrontierMath, tier 4Not measured72.5%Only GPT-5.5 measured
ProofBenchNot measured50.0%Only GPT-5.5 measured
LMArena Text: math14631487GPT-5.5
LMArena Text: mathematical work14651496GPT-5.5
Neither model has a result on the other 7 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index08
Knowledge benchmarks for Gemma 4 31B and GPT-5.5
BenchmarkGemma 4 31BGPT-5.5Leader
MMLU ProNot measured88.1%Only GPT-5.5 measured
LegalBench (Vals AI)Not measured86.5%Only GPT-5.5 measured
CaseLaw (Vals AI)52.6%66.2%GPT-5.5
CorpFin (Vals AI)Not measured68.4%Only GPT-5.5 measured
TaxEval (Vals AI)Not measured75.0%Only GPT-5.5 measured
MortgageTax (Vals AI)61.4%68.8%GPT-5.5
MedCode (Vals AI)Not measured49.1%Only GPT-5.5 measured
SimpleQA Verified10.4%63.0%GPT-5.5
LMArena Text: science14571485GPT-5.5
LMArena Text: medicine and healthcare14511465GPT-5.5
LMArena Text: legal and government14381491GPT-5.5
LMArena Text: business and finance14301478GPT-5.5
FACTS Benchmark Suite45.3%66.7%GPT-5.5
Neither model has a result on the other 5 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index01
Web design benchmarks for Gemma 4 31B and GPT-5.5
BenchmarkGemma 4 31BGPT-5.5Leader
Design Arena: websitesNot measured1256Only GPT-5.5 measured
Design Arena: UI componentsNot measured1257Only GPT-5.5 measured
Design Arena: game devNot measured1293Only GPT-5.5 measured
Design Arena: data vizNot measured1264Only GPT-5.5 measured
Design Arena: 3DNot measured1197Only GPT-5.5 measured
Design Arena: all code categoriesNot measured1260Only GPT-5.5 measured
Design Arena: SVGNot measured1234Only GPT-5.5 measured
Design Arena: ASCII artNot measured1255Only GPT-5.5 measured
LiveBench design58.4%70.3%GPT-5.5
GeneralLike-for-likePublic boards, shown as published, kept out of the index128
General benchmarks for Gemma 4 31B and GPT-5.5
BenchmarkGemma 4 31BGPT-5.5Leader
LiveBench data analysis53.3%81.6%GPT-5.5
Public Benefits Bench (Vals AI)Not measured60.9%Only GPT-5.5 measured
SAGE (Vals AI)55.0%51.5%Gemma 4 31B
MedScribe (Vals AI)Not measured86.9%Only GPT-5.5 measured
Vals Multimodal IndexNot measured68.1%Only GPT-5.5 measured
IFBench75.6%75.9%GPT-5.5
Long context reasoning (AA-LCR)69.7%84.3%GPT-5.5
DTBench82.7%96.0%GPT-5.5
LMCA39.3%54.3%GPT-5.5
Surface Evolver Bench30.6%88.1%GPT-5.5
Furniture assemblyNot measured44.2%Only GPT-5.5 measured
ExploitBenchNot measured47.4%Only GPT-5.5 measured
CL-bench LifeNot measured22.2%Only GPT-5.5 measured
LMArena Text14431471GPT-5.5
LMArena Text: instruction following14331478GPT-5.5
LMArena Text: creative writing14171455GPT-5.5
LMArena Vision12761297GPT-5.5
LMArena Text: English14511470GPT-5.5
LMArena Text: non-English14301467GPT-5.5
LMArena Text: Chinese14751534GPT-5.5
LMArena Text: GermanNot measured1480Only GPT-5.5 measured
LMArena Text: JapaneseNot measured1499Only GPT-5.5 measured
LMArena Text: KoreanNot measured1460Only GPT-5.5 measured
LMArena Text: French14361485GPT-5.5
LMArena Text: Spanish14421468GPT-5.5
LMArena Text: Russian14601473GPT-5.5
LMArena Text: Polish14331465GPT-5.5
LMArena Text: multi-turn14521476GPT-5.5
LMArena Text: longer queries14461484GPT-5.5
LMArena Text: writing and language14321466GPT-5.5
LMArena Text: entertainment and media14051449GPT-5.5
LMArena Vision: English12711292GPT-5.5
LMArena Vision: Chinese13261360GPT-5.5
LMArena Vision: reading text12821306GPT-5.5
LMArena Vision: humor12801293GPT-5.5
LMArena Vision: captioning1218Not measuredOnly Gemma 4 31B measured
LMArena Vision: recognizing things12261231GPT-5.5
LMArena Vision: creative writing12851301GPT-5.5
LMArena Document14251486GPT-5.5
Neither model has a result on the other 7 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-5.5
BenchmarkGemma 4 31BGPT-5.5Leader
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$0.0003Fits in one request
GPT-5.5$0.020Fits in one request

Gemma 4 31B costs less

Repository review

50K input + 3K output tokens

Gemma 4 31B$0.0082Fits in one request
GPT-5.5$0.340Fits in one request

Gemma 4 31B costs less

Agent loop

200K input + 10K output tokens

Gemma 4 31B$0.032Fits in one request
GPT-5.5$1.30Fits in one request

Gemma 4 31B 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-5.5
SpecificationGoogle Gemma 4 31BOpenAI GPT-5.5
MakerGoogleOpenAI
WeightsOpen weightsNo public weights listed
ReleasedApril 2, 2026April 24, 2026
Context window262K1.05M
Input $/Mtok$0.14$5.00
Output $/Mtok$0.40$30.00
Coding index39.071.3
Independent measurements behind it25
Cost per solved taskNot computable$1.65
SourceOpenRouter listingOpenRouter listing

Questions

Is Gemma 4 31B better than GPT-5.5 for coding?

Pick GPT-5.5 for coding: it scores 71.3 on the Vibe Coding Index against 39.0 for Gemma 4 31B. Pick Gemma 4 31B for output tokens 75.0x cheaper ($0.40 vs $30.00 per million) and work measured by SAGE (Vals AI), where it leads.

Which is cheaper, Gemma 4 31B or GPT-5.5?

Gemma 4 31B costs $0.14 per million input tokens and $0.40 per million output tokens; GPT-5.5 costs $5.00 and $30.00. Cost per solved task is known for GPT-5.5 only.

Which has the bigger context window, Gemma 4 31B or GPT-5.5?

GPT-5.5, with 1.05M tokens against 262K.

How many benchmarks is the Gemma 4 31B vs GPT-5.5 comparison based on?

69 benchmarks have a result for both models as of October 22, 2026. Gemma 4 31B's coding index rests on 2 independent benchmarks and GPT-5.5's on 5. Where only one model has a result, the page says so instead of naming a winner.

Compare next

Embed this comparison on your site

Free under CC BY 4.0, while the source link stays visible. Preview the frame · Terms and other embeds

<iframe src="https://llmbenchmark.ai/embed/compare/gemma-4-31b-it-vs-gpt-5-5" title="Gemma 4 31B vs GPT-5.5, LLMBenchmark.ai" width="100%" height="360" style="border:0;max-width:100%" loading="lazy"></iframe>
<p style="font:13px/1.4 sans-serif;margin:6px 0 0">Source: <a href="https://llmbenchmark.ai/compare/gemma-4-31b-it-vs-gpt-5-5">LLMBenchmark.ai, Gemma 4 31B vs GPT-5.5</a></p>