Enter opens the highlighted result. Esc closes.

Gemma 4 31B vs GPT-5.6 Terra

GPT-5.6 Terra leads Gemma 4 31B on coding, 63.9 to 39.0 on the Vibe Coding Index, and Gemma 4 31B costs 30.0x less per output token ($0.40 vs $12.00 per million). Updated October 22, 2026; 68 benchmarks have a result for both.

Replace Gemma 4 31B Replace GPT-5.6 Terra
Gemma 4 31BGoogle
Coding
39.0
Overall
42.1
Output $/M
$0.40
13
GPT-5.6 TerraOpenAI
Coding
63.9
Overall
68.1
Output $/M
$12.00
55
vs boards won of 68 shared

Gemma 4 31B or GPT-5.6 Terra: which to pick

Pick GPT-5.6 Terra for coding: it scores 63.9 on the Vibe Coding Index against 39.0 for Gemma 4 31B.

Pick Gemma 4 31B for output tokens 30.0x cheaper ($0.40 vs $12.00 per million) and work measured by Kagi LLM Benchmark or LMArena Text: science, where it leads.

Model AGemma 4 31BGoogle39.0/1002 independent benchmarksNo cost per solved taskModel BGPT-5.6 TerraOpenAI63.9/1006 independent benchmarks$0.42 per solved task
Shared results
68
Gemma 4 31B only
3
GPT-5.6 Terra only
59
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-5.6 Terra 63.9 from 6 benchmarks.
Widest category
Agentic work, 38.2 points apart on average over 5 shared benchmarks.
Widest single gap
LiveBench browser: Gemma 4 31B 4.0% against GPT-5.6 Terra 84.0%, 80.0 points apart on a 0 to 100 scale.
Output price
Gemma 4 31B $0.40 against GPT-5.6 Terra $12.00 per million output tokens, 30.0x apart.
Context window
Gemma 4 31B takes 262K tokens, GPT-5.6 Terra 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-5.6 TerraTies
Coding012
Intelligence02
Agentic work05
Reasoning19
Math03
Knowledge25
Web design01
General1018
All shared boards1355
CodingLike-for-likeSix of these boards feed the Vibe Coding Index012
Coding benchmarks for Gemma 4 31B and GPT-5.6 Terra
BenchmarkGemma 4 31BGPT-5.6 TerraLeader
LiveBench coding49.0%86.2%GPT-5.6 Terra
LMArena WebDev13671522GPT-5.6 Terra
AA Coding Index43.476.7GPT-5.6 Terra
LiveCodeBench (Vals AI)Not measured85.9%Only GPT-5.6 Terra measured
SWE-bench Verified (Vals AI)Not measured95.4%Only GPT-5.6 Terra measured
Terminal-Bench 4 (Vals AI)Not measured22.7%Only GPT-5.6 Terra measured
Vibe Code Bench (Vals AI)Not measured74.6%Only GPT-5.6 Terra measured
Code migration (Vals AI)Not measured47.8%Only GPT-5.6 Terra measured
ProgramBench (Vals AI)Not measured0.5%Only GPT-5.6 Terra measured
IOI (Vals AI)Not measured87.6%Only GPT-5.6 Terra measured
Terminal-Bench 2.1 (Vals AI)Not measured77.5%Only GPT-5.6 Terra measured
Vibe Code Bench 1-100 (Vals AI)Not measured14.8%Only GPT-5.6 Terra measured
SkillsBench (Vals AI)Not measured58.9%Only GPT-5.6 Terra measured
CyberBench (Vals AI)Not measured72.1%Only GPT-5.6 Terra measured
SciCode45.5%55.0%GPT-5.6 Terra
Terminal-Bench Hard (Artificial Analysis run)36.4%62.9%GPT-5.6 Terra
FrontierCodeNot measured41.3%Only GPT-5.6 Terra measured
WeirdML52.3%78.3%GPT-5.6 Terra
ALE-Bench925.51951.4GPT-5.6 Terra
LMArena Text: coding14591485GPT-5.6 Terra
LMArena Text: software and IT14631482GPT-5.6 Terra
LMArena WebDev: React13591513GPT-5.6 Terra
LMArena WebDev: HTML13801526GPT-5.6 Terra
LMArena WebDev: from an imageNot measured1541Only GPT-5.6 Terra measured
SciCode8.9%10.8%GPT-5.6 Terra
Neither model has a result on the other 14 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
Intelligence benchmarks for Gemma 4 31B and GPT-5.6 Terra
BenchmarkGemma 4 31BGPT-5.6 TerraLeader
AA Intelligence Index14.742.1GPT-5.6 Terra
Epoch Capabilities Index143.0160.1GPT-5.6 Terra
Agentic workLike-for-likePublic boards, shown as published, kept out of the index05
Agentic work benchmarks for Gemma 4 31B and GPT-5.6 Terra
BenchmarkGemma 4 31BGPT-5.6 TerraLeader
AA Agentic Index4.243.2GPT-5.6 Terra
Design Arena: agents building web appsNot measured1120Only GPT-5.6 Terra measured
Design Arena: agents building full-stack appsNot measured1061Only GPT-5.6 Terra measured
Design Arena: agents building mobile appsNot measured1111Only GPT-5.6 Terra measured
Design Arena: agents building Android appsNot measured1062Only GPT-5.6 Terra measured
Design Arena: agents building gamesNot measured1083Only GPT-5.6 Terra measured
Design Arena: HTML slidesNot measured1058Only GPT-5.6 Terra measured
Design Arena: PowerPoint slidesNot measured1105Only GPT-5.6 Terra measured
LiveBench research44.4%60.4%GPT-5.6 Terra
LiveBench browser4.0%84.0%GPT-5.6 Terra
LiveBench long agentic13.3%48.4%GPT-5.6 Terra
SREBench (Vals AI)Not measured9.5%Only GPT-5.6 Terra measured
Finance Agent (Vals AI)Not measured54.4%Only GPT-5.6 Terra measured
Tax Agent Bench (Vals AI)Not measured30.0%Only GPT-5.6 Terra measured
Legal Research Bench (Vals AI)Not measured41.4%Only GPT-5.6 Terra measured
Excel Modeling Benchmark (Vals AI)Not measured66.2%Only GPT-5.6 Terra measured
Terminal-Bench-Science (Vals AI)Not measured10.0%Only GPT-5.6 Terra measured
τ²-Bench Telecom65.5%86.3%GPT-5.6 Terra
APEX-AgentsNot measured58.2%Only GPT-5.6 Terra measured
BALROGNot measured53.2%Only GPT-5.6 Terra measured
Vending-Bench 2Not measured$7,343Only GPT-5.6 Terra measured
Neither model has a result on the other 20 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index19
Reasoning benchmarks for Gemma 4 31B and GPT-5.6 Terra
BenchmarkGemma 4 31BGPT-5.6 TerraLeader
GPQA DiamondNot measured90.9%Only GPT-5.6 Terra measured
LiveBench reasoning68.3%90.6%GPT-5.6 Terra
MMMU Pro (Vals AI)Not measured86.5%Only GPT-5.6 Terra measured
GPQA Diamond (Artificial Analysis run)85.7%92.5%GPT-5.6 Terra
Humanity's Last Exam (Artificial Analysis run)23.6%42.9%GPT-5.6 Terra
GPQA Diamond (Epoch AI run)75.8%93.3%GPT-5.6 Terra
ARC-AGINot measured96.5%Only GPT-5.6 Terra measured
ARC-AGI-2Not measured83.9%Only GPT-5.6 Terra measured
SimpleBenchNot measured48.9%Only GPT-5.6 Terra measured
Chess puzzlesNot measured47.0%Only GPT-5.6 Terra measured
Mystery game puzzlesNot measured35.0%Only GPT-5.6 Terra measured
LMArena Text: hard prompts14481469GPT-5.6 Terra
LMArena Text: expert14641496GPT-5.6 Terra
LMArena Text: hard prompts in English14561465GPT-5.6 Terra
LMArena Vision: diagrams12831295GPT-5.6 Terra
LMArena Vision: homework13101312GPT-5.6 Terra
Kagi LLM Benchmark63.5%51.3%Gemma 4 31B
Neither model has a result on the other 5 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index03
Math benchmarks for Gemma 4 31B and GPT-5.6 Terra
BenchmarkGemma 4 31BGPT-5.6 TerraLeader
OTIS Mock AIME73.3%99.7%GPT-5.6 Terra
FrontierMath, tiers 1 to 3Not measured86.0%Only GPT-5.6 Terra measured
FrontierMath, tier 4Not measured70.7%Only GPT-5.6 Terra measured
ProofBenchNot measured74.0%Only GPT-5.6 Terra measured
LMArena Text: math14631465GPT-5.6 Terra
LMArena Text: mathematical work14651478GPT-5.6 Terra
Neither model has a result on the other 7 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index25
Knowledge benchmarks for Gemma 4 31B and GPT-5.6 Terra
BenchmarkGemma 4 31BGPT-5.6 TerraLeader
MMLU ProNot measured86.7%Only GPT-5.6 Terra measured
LegalBench (Vals AI)Not measured85.1%Only GPT-5.6 Terra measured
CaseLaw (Vals AI)52.6%Not measuredOnly Gemma 4 31B measured
CorpFin (Vals AI)Not measured65.3%Only GPT-5.6 Terra measured
TaxEval (Vals AI)Not measured76.2%Only GPT-5.6 Terra measured
MortgageTax (Vals AI)61.4%67.3%GPT-5.6 Terra
MedCode (Vals AI)Not measured43.4%Only GPT-5.6 Terra measured
SimpleQA Verified10.4%43.2%GPT-5.6 Terra
LMArena Text: science14571452Gemma 4 31B
LMArena Text: medicine and healthcare14511430Gemma 4 31B
LMArena Text: legal and government14381461GPT-5.6 Terra
LMArena Text: business and finance14301448GPT-5.6 Terra
FACTS Benchmark Suite45.3%52.0%GPT-5.6 Terra
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-5.6 Terra
BenchmarkGemma 4 31BGPT-5.6 TerraLeader
Design Arena: websitesNot measured1268Only GPT-5.6 Terra measured
Design Arena: UI componentsNot measured1256Only GPT-5.6 Terra measured
Design Arena: game devNot measured1229Only GPT-5.6 Terra measured
Design Arena: data vizNot measured1240Only GPT-5.6 Terra measured
Design Arena: 3DNot measured1134Only GPT-5.6 Terra measured
Design Arena: all code categoriesNot measured1241Only GPT-5.6 Terra measured
Design Arena: SVGNot measured1219Only GPT-5.6 Terra measured
Design Arena: ASCII artNot measured1177Only GPT-5.6 Terra measured
LiveBench design58.4%76.0%GPT-5.6 Terra
GeneralLike-for-likePublic boards, shown as published, kept out of the index1018
General benchmarks for Gemma 4 31B and GPT-5.6 Terra
BenchmarkGemma 4 31BGPT-5.6 TerraLeader
LiveBench data analysis53.3%79.3%GPT-5.6 Terra
Public Benefits Bench (Vals AI)Not measured62.4%Only GPT-5.6 Terra measured
SAGE (Vals AI)55.0%47.0%Gemma 4 31B
MedScribe (Vals AI)Not measured82.9%Only GPT-5.6 Terra measured
Vals Multimodal IndexNot measured65.1%Only GPT-5.6 Terra measured
IFBench75.6%71.2%Gemma 4 31B
Long context reasoning (AA-LCR)69.7%83.0%GPT-5.6 Terra
DTBench82.7%93.3%GPT-5.6 Terra
LMCA39.3%54.9%GPT-5.6 Terra
Surface Evolver Bench30.6%83.8%GPT-5.6 Terra
Furniture assemblyNot measured54.2%Only GPT-5.6 Terra measured
LMArena Text14431448GPT-5.6 Terra
LMArena Text: instruction following14331456GPT-5.6 Terra
LMArena Text: creative writing14171413Gemma 4 31B
LMArena Vision12761271Gemma 4 31B
LMArena Text: English14511449Gemma 4 31B
LMArena Text: non-English14301441GPT-5.6 Terra
LMArena Text: Chinese14751514GPT-5.6 Terra
LMArena Text: GermanNot measured1460Only GPT-5.6 Terra measured
LMArena Text: JapaneseNot measured1454Only GPT-5.6 Terra measured
LMArena Text: KoreanNot measured1427Only GPT-5.6 Terra measured
LMArena Text: French14361466GPT-5.6 Terra
LMArena Text: Spanish14421451GPT-5.6 Terra
LMArena Text: Russian14601452Gemma 4 31B
LMArena Text: Polish14331446GPT-5.6 Terra
LMArena Text: multi-turn14521453GPT-5.6 Terra
LMArena Text: longer queries14461454GPT-5.6 Terra
LMArena Text: writing and language14321431Gemma 4 31B
LMArena Text: entertainment and media14051418GPT-5.6 Terra
LMArena Vision: English12711272GPT-5.6 Terra
LMArena Vision: Chinese13261337GPT-5.6 Terra
LMArena Vision: reading text12821278Gemma 4 31B
LMArena Vision: humor12801236Gemma 4 31B
LMArena Vision: captioning1218Not measuredOnly Gemma 4 31B measured
LMArena Vision: recognizing things1226Not measuredOnly Gemma 4 31B measured
LMArena Vision: creative writing12851248Gemma 4 31B
LMArena Document14251472GPT-5.6 Terra
Neither model has a result on the other 9 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-5.6 Terra
BenchmarkGemma 4 31BGPT-5.6 TerraLeader
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-5.6 Terra$0.0080Fits in one request

Gemma 4 31B costs less

Repository review

50K input + 3K output tokens

Gemma 4 31B$0.0082Fits in one request
GPT-5.6 Terra$0.136Fits in one request

Gemma 4 31B costs less

Agent loop

200K input + 10K output tokens

Gemma 4 31B$0.032Fits in one request
GPT-5.6 Terra$0.520Fits 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-5.6 Terra
SpecificationGoogle Gemma 4 31BOpenAI GPT-5.6 Terra
MakerGoogleOpenAI
WeightsOpen weightsNo public weights listed
ReleasedApril 2, 2026July 9, 2026
Context window262K1.05M
Input $/Mtok$0.14$2.00
Output $/Mtok$0.40$12.00
Coding index39.063.9
Independent measurements behind it26
Cost per solved taskNot computable$0.42
SourceOpenRouter listingOpenRouter listing

Questions

Is Gemma 4 31B better than GPT-5.6 Terra for coding?

Pick GPT-5.6 Terra for coding: it scores 63.9 on the Vibe Coding Index against 39.0 for Gemma 4 31B. Pick Gemma 4 31B for output tokens 30.0x cheaper ($0.40 vs $12.00 per million) and work measured by Kagi LLM Benchmark or LMArena Text: science, where it leads.

Which is cheaper, Gemma 4 31B or GPT-5.6 Terra?

Gemma 4 31B costs $0.14 per million input tokens and $0.40 per million output tokens; GPT-5.6 Terra costs $2.00 and $12.00. Cost per solved task is known for GPT-5.6 Terra only.

Which has the bigger context window, Gemma 4 31B or GPT-5.6 Terra?

GPT-5.6 Terra, with 1.05M tokens against 262K.

How many benchmarks is the Gemma 4 31B vs GPT-5.6 Terra comparison based on?

68 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-5.6 Terra's on 6. 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-5-6-terra" title="Gemma 4 31B vs GPT-5.6 Terra, 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-5-6-terra">LLMBenchmark.ai, Gemma 4 31B vs GPT-5.6 Terra</a></p>