LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

Claude Opus 4.7 vs Gemini 3.1 Pro Preview

Updated October 21, 2026. 97 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace Claude Opus 4.7 Replace Gemini 3.1 Pro Preview
Claude Opus 4.7Anthropic
Coding
51.1
Overall
40.7
Output $/M
$25.00
68
Gemini 3.1 Pro PreviewGoogle
Coding
35.3
Overall
29.7
Output $/M
$12.00
27
vs boards won of 97 shared, 2 tied

Decision reading

Pick Claude Opus 4.7 for coding: it scores 51.1 on the Vibe Coding Index against 35.3 for Gemini 3.1 Pro Preview.

Pick Gemini 3.1 Pro Preview when output tokens cost 2.1x less, or when a solved task costs $0.99 against $2.95.

Model AClaude Opus 4.7Anthropic51.1/1001 independent benchmark$2.95 per solved taskModel BGemini 3.1 Pro PreviewGoogle35.3/1001 independent benchmark$0.99 per solved task
Shared results
97
Claude Opus 4.7 only
4
Gemini 3.1 Pro Preview only
34
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
Claude Opus 4.7 51.1 from 1 benchmark, Gemini 3.1 Pro Preview 35.3 from 1 benchmark.
Widest category
Agentic work, 19.2 points apart on average over 9 shared benchmarks.
Widest single gap
Vending-Bench 2: Claude Opus 4.7 $10,937 against Gemini 3.1 Pro Preview $911, 62.7 points apart on a 0 to 100 scale.
Output price
Claude Opus 4.7 $25.00 against Gemini 3.1 Pro Preview $12.00 per million output tokens, 2.1x apart.
Context window
Claude Opus 4.7 takes 1M tokens, Gemini 3.1 Pro Preview 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
CategoryClaude Opus 4.7BarsGemini 3.1 Pro PreviewTies
Coding1422 ties
Intelligence10
Agentic work81
Reasoning610
Math61
Knowledge84
General259
All shared boards68272 ties
CodingLike-for-likeFour of these boards feed the Vibe Coding Index142
Coding benchmarks for Claude Opus 4.7 and Gemini 3.1 Pro Preview
BenchmarkClaude Opus 4.7Gemini 3.1 Pro PreviewLeader
LMArena WebDev15581447Claude Opus 4.7
AA Coding Index73.668.8Claude Opus 4.7
LiveCodeBench (Vals AI)85.1%88.5%Gemini 3.1 Pro Preview
SWE-bench Verified (Vals AI)82.0%78.8%Claude Opus 4.7
Terminal-Bench 4 (Vals AI)Not measured2.5%Only Gemini 3.1 Pro Preview measured
Vibe Code Bench (Vals AI)71.0%32.0%Claude Opus 4.7
Code migration (Vals AI)43.9%17.3%Claude Opus 4.7
ProgramBench (Vals AI)0.0%0.0%Tie
IOI (Vals AI)Not measured51.8%Only Gemini 3.1 Pro Preview measured
Terminal-Bench 2.1 (Vals AI)68.5%70.8%Gemini 3.1 Pro Preview
Vibe Code Bench 1-100 (Vals AI)Not measured6.7%Only Gemini 3.1 Pro Preview measured
SciCodeNot measured58.7%Only Gemini 3.1 Pro Preview measured
Terminal-Bench Hard (Artificial Analysis run)54.5%53.8%Claude Opus 4.7
Terminal-Bench80.2%80.2%Tie
FrontierCode38.5%Not measuredOnly Claude Opus 4.7 measured
GSO-Bench44.1%22.6%Claude Opus 4.7
WeirdML76.4%72.1%Claude Opus 4.7
MirrorCode31.1%8.9%Claude Opus 4.7
ALE-BenchNot measured1160.6Only Gemini 3.1 Pro Preview measured
LMArena Text: coding15191484Claude Opus 4.7
LMArena Text: software and IT15161487Claude Opus 4.7
LMArena WebDev: React15601438Claude Opus 4.7
LMArena WebDev: HTML15541490Claude Opus 4.7
LMArena WebDev: from an image15771481Claude Opus 4.7
Neither model has a result on the other 14 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
Intelligence benchmarks for Claude Opus 4.7 and Gemini 3.1 Pro Preview
BenchmarkClaude Opus 4.7Gemini 3.1 Pro PreviewLeader
AA Intelligence Index40.729.7Claude Opus 4.7
Epoch Capabilities Index156.8Not measuredOnly Claude Opus 4.7 measured
Agentic workLike-for-likePublic boards, shown as published, kept out of the index81
Agentic work benchmarks for Claude Opus 4.7 and Gemini 3.1 Pro Preview
BenchmarkClaude Opus 4.7Gemini 3.1 Pro PreviewLeader
AA Agentic Index38.68.2Claude Opus 4.7
Design Arena: agents building web appsNot measured1125Only Gemini 3.1 Pro Preview measured
Design Arena: agents building full-stack appsNot measured1068Only Gemini 3.1 Pro Preview measured
Design Arena: agents building mobile appsNot measured1094Only Gemini 3.1 Pro Preview measured
Design Arena: agents building Android appsNot measured1055Only Gemini 3.1 Pro Preview measured
Design Arena: agents building Godot gamesNot measured1236Only Gemini 3.1 Pro Preview measured
Design Arena: agents building gamesNot measured1114Only Gemini 3.1 Pro Preview measured
Design Arena: HTML slidesNot measured1153Only Gemini 3.1 Pro Preview measured
Design Arena: PowerPoint slidesNot measured1064Only Gemini 3.1 Pro Preview measured
SREBench (Vals AI)Not measured0.0%Only Gemini 3.1 Pro Preview measured
Finance Agent (Vals AI)51.5%43.0%Claude Opus 4.7
Tax Agent Bench (Vals AI)23.7%9.3%Claude Opus 4.7
Legal Research Bench (Vals AI)38.5%20.7%Claude Opus 4.7
Excel Modeling Benchmark (Vals AI)63.7%52.6%Claude Opus 4.7
BioMysteryBench (Vals AI)Not measured60.0%Only Gemini 3.1 Pro Preview measured
Terminal-Bench-Science (Vals AI)Not measured1.4%Only Gemini 3.1 Pro Preview measured
Vals Science IndexNot measured27.4%Only Gemini 3.1 Pro Preview measured
τ²-Bench Telecom88.6%95.6%Gemini 3.1 Pro Preview
APEX-Agents49.2%35.1%Claude Opus 4.7
METR time horizonsNot measured77.0%Only Gemini 3.1 Pro Preview measured
OSWorld 2.018.2%Not measuredOnly Claude Opus 4.7 measured
BALROGNot measured57.0%Only Gemini 3.1 Pro Preview measured
DeepResearch BenchNot measured47.8%Only Gemini 3.1 Pro Preview measured
PostTrainBench28.6%22.0%Claude Opus 4.7
Vending-Bench 2$10,937$911Claude Opus 4.7
LMArena Search1230Not measuredOnly Claude Opus 4.7 measured
Neither model has a result on the other 13 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index610
Reasoning benchmarks for Claude Opus 4.7 and Gemini 3.1 Pro Preview
BenchmarkClaude Opus 4.7Gemini 3.1 Pro PreviewLeader
GPQA Diamond90.2%95.5%Gemini 3.1 Pro Preview
MysteryMechanism (Vals AI)Not measured20.7%Only Gemini 3.1 Pro Preview measured
MMMU Pro (Vals AI)85.5%88.2%Gemini 3.1 Pro Preview
GPQA Diamond (Artificial Analysis run)91.4%94.1%Gemini 3.1 Pro Preview
Humanity's Last Exam (Artificial Analysis run)42.3%47.0%Gemini 3.1 Pro Preview
GPQA Diamond (Epoch AI run)90.2%94.4%Gemini 3.1 Pro Preview
ARC-AGI93.5%98.0%Gemini 3.1 Pro Preview
ARC-AGI-275.8%77.1%Gemini 3.1 Pro Preview
SimpleBench61.7%79.6%Gemini 3.1 Pro Preview
Chess puzzles41.0%51.0%Gemini 3.1 Pro Preview
Mystery game puzzles28.0%34.0%Gemini 3.1 Pro Preview
EBR-bench19.0%14.3%Claude Opus 4.7
LMArena Text: hard prompts15061485Claude Opus 4.7
LMArena Text: expert15211483Claude Opus 4.7
LMArena Text: hard prompts in English15081480Claude Opus 4.7
LMArena Vision: diagrams13361311Claude Opus 4.7
LMArena Vision: homework13371316Claude Opus 4.7
Neither model has a result on the other 4 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index61
Math benchmarks for Claude Opus 4.7 and Gemini 3.1 Pro Preview
BenchmarkClaude Opus 4.7Gemini 3.1 Pro PreviewLeader
AIME96.3%98.1%Gemini 3.1 Pro Preview
OTIS Mock AIME97.8%95.6%Claude Opus 4.7
FrontierMath, tiers 1 to 370.2%59.6%Claude Opus 4.7
FrontierMath, tier 431.7%26.8%Claude Opus 4.7
ProofBench54.0%26.0%Claude Opus 4.7
LMArena Text: math14981486Claude Opus 4.7
LMArena Text: mathematical work15091478Claude Opus 4.7
Neither model has a result on the other 6 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index84
Knowledge benchmarks for Claude Opus 4.7 and Gemini 3.1 Pro Preview
BenchmarkClaude Opus 4.7Gemini 3.1 Pro PreviewLeader
MMLU Pro89.9%91.0%Gemini 3.1 Pro Preview
LegalBench (Vals AI)85.3%87.4%Gemini 3.1 Pro Preview
CaseLaw (Vals AI)68.4%64.8%Claude Opus 4.7
CorpFin (Vals AI)66.1%64.5%Claude Opus 4.7
TaxEval (Vals AI)75.3%72.9%Claude Opus 4.7
MortgageTax (Vals AI)70.3%69.4%Claude Opus 4.7
MedQA (Vals AI)Not measured96.4%Only Gemini 3.1 Pro Preview measured
MedCode (Vals AI)54.9%59.1%Gemini 3.1 Pro Preview
SimpleQA Verified51.7%73.5%Gemini 3.1 Pro Preview
LMArena Text: science15081497Claude Opus 4.7
LMArena Text: medicine and healthcare14951486Claude Opus 4.7
LMArena Text: legal and government15041487Claude Opus 4.7
LMArena Text: business and finance14901459Claude Opus 4.7
Neither model has a result on the other 4 boards here
Web designDirectional onlyPublic boards, shown as published, kept out of the index8 boards, none shared
Web design benchmarks for Claude Opus 4.7 and Gemini 3.1 Pro Preview
BenchmarkClaude Opus 4.7Gemini 3.1 Pro PreviewLeader
Design Arena: websitesNot measured1234Only Gemini 3.1 Pro Preview measured
Design Arena: UI componentsNot measured1226Only Gemini 3.1 Pro Preview measured
Design Arena: game devNot measured1203Only Gemini 3.1 Pro Preview measured
Design Arena: data vizNot measured1209Only Gemini 3.1 Pro Preview measured
Design Arena: 3DNot measured1234Only Gemini 3.1 Pro Preview measured
Design Arena: all code categoriesNot measured1231Only Gemini 3.1 Pro Preview measured
Design Arena: SVGNot measured1285Only Gemini 3.1 Pro Preview measured
Design Arena: ASCII artNot measured1270Only Gemini 3.1 Pro Preview measured
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index259
General benchmarks for Claude Opus 4.7 and Gemini 3.1 Pro Preview
BenchmarkClaude Opus 4.7Gemini 3.1 Pro PreviewLeader
Public Benefits Bench (Vals AI)Not measured53.8%Only Gemini 3.1 Pro Preview measured
SAGE (Vals AI)56.1%48.7%Claude Opus 4.7
MedScribe (Vals AI)83.0%76.1%Claude Opus 4.7
Vals Multimodal Index67.4%56.1%Claude Opus 4.7
IFBench58.6%77.1%Gemini 3.1 Pro Preview
Long context reasoning (AA-LCR)78.7%82.0%Gemini 3.1 Pro Preview
DTBench94.7%97.1%Gemini 3.1 Pro Preview
LMCA52.2%53.8%Gemini 3.1 Pro Preview
CL-benchNot measured20.8%Only Gemini 3.1 Pro Preview measured
Furniture assembly33.3%26.7%Claude Opus 4.7
ExploitBench26.5%26.1%Claude Opus 4.7
CL-bench LifeNot measured16.9%Only Gemini 3.1 Pro Preview measured
LMArena Text14901480Claude Opus 4.7
LMArena Text: instruction following14981466Claude Opus 4.7
LMArena Text: creative writing14861482Claude Opus 4.7
LMArena Vision13171296Claude Opus 4.7
LMArena Text: English14961478Claude Opus 4.7
LMArena Text: non-English14801476Claude Opus 4.7
LMArena Text: Chinese15311527Claude Opus 4.7
LMArena Text: German14951490Claude Opus 4.7
LMArena Text: Japanese14741491Gemini 3.1 Pro Preview
LMArena Text: Korean14641455Claude Opus 4.7
LMArena Text: French15031487Claude Opus 4.7
LMArena Text: Spanish14941481Claude Opus 4.7
LMArena Text: Russian14941497Gemini 3.1 Pro Preview
LMArena Text: Polish14891498Gemini 3.1 Pro Preview
LMArena Text: multi-turn15051487Claude Opus 4.7
LMArena Text: longer queries15051482Claude Opus 4.7
LMArena Text: writing and language14921481Claude Opus 4.7
LMArena Text: entertainment and media14741465Claude Opus 4.7
LMArena Vision: English13221287Claude Opus 4.7
LMArena Vision: Chinese13711364Claude Opus 4.7
LMArena Vision: reading text13271304Claude Opus 4.7
LMArena Vision: humor13091322Gemini 3.1 Pro Preview
LMArena Vision: captioningNot measured1345Only Gemini 3.1 Pro Preview measured
LMArena Vision: recognizing things12781330Gemini 3.1 Pro Preview
LMArena Vision: creative writing13261324Claude Opus 4.7
LMArena Document14951444Claude Opus 4.7
Neither model has a result on the other 8 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for Claude Opus 4.7 and Gemini 3.1 Pro Preview
BenchmarkClaude Opus 4.7Gemini 3.1 Pro PreviewLeader
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

Claude Opus 4.7$0.018Fits in one request
Gemini 3.1 Pro Preview$0.0080Fits in one request

Gemini 3.1 Pro Preview costs less

Repository review

50K input + 3K output tokens

Claude Opus 4.7$0.325Fits in one request
Gemini 3.1 Pro Preview$0.136Fits in one request

Gemini 3.1 Pro Preview costs less

Agent loop

200K input + 10K output tokens

Claude Opus 4.7$1.25Fits in one request
Gemini 3.1 Pro Preview$0.520Fits in one request

Gemini 3.1 Pro Preview 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 Claude Opus 4.7 and Gemini 3.1 Pro Preview
SpecificationAnthropic Claude Opus 4.7Google Gemini 3.1 Pro Preview
MakerAnthropicGoogle
WeightsNo public weights listedNo public weights listed
ReleasedApril 16, 2026February 19, 2026
Context window1M1.05M
Input $/Mtok$5.00$2.00
Output $/Mtok$25.00$12.00
Coding index51.135.3
Independent measurements behind it11
Cost per solved task$2.95$0.99
SourceOpenRouter listingOpenRouter listing

Questions

Is Claude Opus 4.7 better than Gemini 3.1 Pro Preview for coding?

Pick Claude Opus 4.7 for coding: it scores 51.1 on the Vibe Coding Index against 35.3 for Gemini 3.1 Pro Preview. Pick Gemini 3.1 Pro Preview when output tokens cost 2.1x less, or when a solved task costs $0.99 against $2.95.

Which is cheaper, Claude Opus 4.7 or Gemini 3.1 Pro Preview?

Claude Opus 4.7 costs $5.00 per million input tokens and $25.00 per million output tokens; Gemini 3.1 Pro Preview costs $2.00 and $12.00. Per solved SWE-bench Verified task, Claude Opus 4.7 costs $2.95 and Gemini 3.1 Pro Preview $0.99.

Which has the bigger context window, Claude Opus 4.7 or Gemini 3.1 Pro Preview?

Gemini 3.1 Pro Preview, with 1.05M tokens against 1M.

How many benchmarks is the Claude Opus 4.7 vs Gemini 3.1 Pro Preview comparison based on?

97 benchmarks have a result for both models as of October 21, 2026. Claude Opus 4.7's coding index rests on 1 independent benchmark and Gemini 3.1 Pro Preview's on 1. Where only one model has a result, the page says so instead of naming a winner.

Compare next