GoogleDeepSeekLLM matchup

Gemini 3.1 Pro Previewvs.DeepSeek-V4.1-Flash

Which one is better?

DeepSeek-V4.1-Flash performs better than Gemini 3.1 Pro Preview and is cheaper: Gemini 3.1 Pro Preview costs 13.3 times (input) and 20 times (output) as much.

DeepSeek-V4.1-Flash ahead

IntelligenceOverall index, 100 points = average

90.1

Gemini 3.1 Pro Preview

Rank 41

99.7

DeepSeek-V4.1-Flash

Rank 29

DeepSeek-V4.1-Flash faster

SpeedTokens per second

122

Gemini 3.1 Pro Preview

176

DeepSeek-V4.1-Flash

DeepSeek-V4.1-Flash cheaper

Price$ per 1M tokens

$4.50

Gemini 3.1 Pro Preview

$0.26

DeepSeek-V4.1-Flash

Gemini 3.1 Pro Preview more popular

Popularity0 to 100

Gemini 3.1 Pro Preview

Rank 110 of 876

DeepSeek-V4.1-Flash

Rank 312 of 876

Above 200,000 tokens of context, Gemini 3.1 Pro Preview costs more: $4 input and $18 output per 1M tokens.

Choose Gemini 3.1 Pro Preview for

No clear advantage

Choose DeepSeek-V4.1-Flash for

FinanceCodingAgentsLegalEducationMathematicsProductivityGeneral

Where is the difference?

Strengths at a glance

Gemini 3.1 Pro PreviewDeepSeek-V4.1-FlashMedian of all ranked models

CodingDeepSeek-V4.1-Flash ahead

Gemini 3.1 Pro Preview81.3Rank 59

DeepSeek-V4.1-Flash98.8Rank 30

Median of ranked models84.1

95 % range: Gemini 3.1 Pro Preview 74.9 to 87.7, DeepSeek-V4.1-Flash 91.4 to 106.3.

AreaGemini 3.1 Pro PreviewDeepSeek-V4.1-Flash

The same values and verdicts as the bars below. In the close-up the net starts at 75 points, just below the weaker model's lowest value; the reference group average is 100. Not in the net because a model has no own value there or the data are too thin: Reasoning, Knowledge.

DeepSeek-V4.1-Flash is ahead in 6 of 10 areas, Gemini 3.1 Pro Preview in none.

level: 2too little data: 2DeepSeek-V4.1-Flash ahead: 6

Areas in the overall index

CodingDeepSeek-V4.1-Flash +1881.3 to 98.8

AgentsDeepSeek-V4.1-Flash +1789.7 to 106.4

MathematicsDeepSeek-V4.1-Flash +1186.7 to 97.3

GeneralDeepSeek-V4.1-Flash +691.5 to 97.1

Multimodallevel95.7 to 95.7

Reasoningonly Gemini 3.1 Pro Preview has a score

Knowledgeonly Gemini 3.1 Pro Preview has a score

Specialist fields

FinanceDeepSeek-V4.1-Flash +2282.1 to 103.7

LegalDeepSeek-V4.1-Flash +1586.1 to 100.8

Healthlevel92.8 to 95.5

Bar length shows the gap in index points (full width 15 points, one standard deviation of the reference group). Solid means clearly ahead, hatched not certain yet, a thin grey bar level.

Show 4 more areas that do not count toward the overall index

EducationDeepSeek-V4.1-Flash +1484.3 to 97.9

ProductivityDeepSeek-V4.1-Flash +891.7 to 99.7

Cybersecurityonly DeepSeek-V4.1-Flash has a score

Searchonly Gemini 3.1 Pro Preview has a score

Bar length shows the gap in index points (full width 15 points, one standard deviation of the reference group). Solid means clearly ahead, hatched not certain yet, a thin grey bar level.

Which model for which task?

Everyday tasks, translated from the area values above. Each row names the values it rests on.

TaskBetter choicebased on

Writing and debugging codeDeepSeek-V4.1-FlashCoding 81.3 to 98.8

Maths and data analysisDeepSeek-V4.1-FlashMathematics 86.7 to 97.3, Reasoning without a score for DeepSeek-V4.1-Flash

Facts and researchopenKnowledge without a score for DeepSeek-V4.1-Flash, Search without a score for DeepSeek-V4.1-Flash

Agents and workflowsDeepSeek-V4.1-FlashAgents 89.7 to 106.4

Fast chat repliesDeepSeek-V4.1-Flash122 to 176 tokens per second

Security analysisopenCybersecurity without a score for Gemini 3.1 Pro Preview

Many requests on a small budgetDeepSeek-V4.1-Flash$4.50 to $0.26 per 1M tokens

Finance, legal and medical textsDeepSeek-V4.1-FlashFinance 82.1 to 103.7, Legal 86.1 to 100.8, Health 92.8 to 95.5

Images, screenshots and PDFslevelMultimodal 95.7 to 95.7

See for yourself: same task, one attempt

The same prompt for both models. We show the latest attempt without fixes; it does not count toward the overall index.

Pelican on a bicycle

Prompt

Generate an SVG of a pelican riding a bicycle

Task origin: Simon Willison

Gemini 3.1 Pro Preview

The response reached the token limit. No complete recording is available.

Oct 6, 2026 · 64 s · $0.10

Not yet visually reviewed. A successful recording does not confirm correct physics or full compliance with the prompt.

Test details

Run date: Oct 6, 2026
API provider: openrouter
Requested API model: google/gemini-3.1-pro-preview
Reasoning setting: high
Token limit including reasoning: 8,192 tokens
Output tokens: 8,188 tokens
Cost of the attempt: $0.0983
Response time: 63.8 s
API model reported in response: google/gemini-3.1-pro-preview
Pinned endpoint: Google AI Studio | google/gemini-3.1-pro-preview-20260219
Temperature: Provider default
Test protocol: community-visual-tests-v1
Technical checks: Failed

Earlier attempts (3)

Oct 2, 2026API request failedReasoning setting: high
Oct 1, 2026API request failedReasoning setting: high
Sep 7, 2026API request failedReasoning setting: high

DeepSeek-V4.1-Flash

The response reached the token limit. No complete recording is available.

Oct 5, 2026 · 26 s · $0.0049

Not yet visually reviewed. A successful recording does not confirm correct physics or full compliance with the prompt.

Test details

Run date: Oct 5, 2026
API provider: openrouter
Requested API model: deepseek/deepseek-v4.1-flash
Reasoning setting: high
Token limit including reasoning: 8,192 tokens
Output tokens: 8,191 tokens
Cost of the attempt: $0.0049
Response time: 25.8 s
API model reported in response: deepseek/deepseek-v4.1-flash
Pinned endpoint: DeepSeek | deepseek/deepseek-v4.1-flash-20260910
Temperature: Provider default
Test protocol: community-visual-tests-v1
Technical checks: Failed

Earlier attempts (2)

Oct 2, 2026API request failedReasoning setting: high
Oct 1, 2026API request failedReasoning setting: high

SVG short test

One SVG task with the same fixed settings for every model, without hidden thinking.

Prompt

Return only one standalone SVG of a pelican riding a bicycle. Use viewBox="0 0 800 600", simple shapes, two clearly visible wheels and a recognizable pelican with a long beak. At most 30 SVG elements and 1800 characters. No Markdown, explanations, comments, scripts, animation, CSS or external resources.

Gemini 3.1 Pro Preview

Unavailable in this round

DeepSeek-V4.1-Flash

Gemini 3.1 Pro Preview and DeepSeek-V4.1-Flash in the overall index

Every model with a place in the overall index, both models of this matchup highlighted. You can add or hide other models.

Overall index (Gradually Meta-Score v4)

20 of 92 models

5 models in this view are measured at a lower reasoning effort than max in the overall index. Dashed column: lower reasoning effort than the top.

Claude Opus 5.5max126.1

Gemini 4 Carbonhigh125.8

Claude Fable 5.1max123.8

Claude Sonnet 5.5max122.8

Gemini 4 Argonhigh122.3

GPT-6 Astramax121.3

GPT-6.1 Solmax119.2

Muse Wonder Fullmax118.3

Claude Opus 5max117.2

Muse Wondermax113.4

Kimi K3.1 Promax112.9

Claude Fable 5max112.5

GPT-6 Solmax110.5

GPT-5.6 Solmax108.7

Muse Spark 1.3max107.4

Gemini 3.7 Flashhigh107.1

DeepSeek V4.1 Flash 1017max106.6

Gemini 3.8 Flashhigh106.5

DeepSeek-V4.1-Flashmax99.7

Gemini 3.1 Pro Previewhigh90.1

Claude Opus 5.5

Gemini 4 Carbon

Claude Fable 5.1

Claude Sonnet 5.5

Gemini 4 Argon

GPT-6 Astra

GPT-6.1 Sol

Muse Wonder Full

Claude Opus 5

Muse Wonder

Kimi K3.1 Pro

Claude Fable 5

GPT-6 Sol

GPT-5.6 Sol

Muse Spark 1.3

Gemini 3.7 Flash

DeepSeek V4.1 Flash 1017

Gemini 3.8 Flash

DeepSeek-V4.1-Flash

Gemini 3.1 Pro Preview

Popularity compared

How often the models are searched for, used and mentioned, on a scale from 0 to 100.

Gemini 3.1 Pro Preview is more popular than DeepSeek-V4.1-Flash: 50 to 38 out of 100.

Over time

Weekly values from October 18, 2026 to November 6, 2026

Gemini 3.1 Pro PreviewDeepSeek-V4.1-Flash

All data side by side

The better value is bold. For the overall index only when the gap is clear.

Metric	Gemini 3.1 Pro Preview	DeepSeek-V4.1-Flash
Intelligence
Overall indexpoints, 100 = reference group average, higher is better	90.1	99.7
Rank	Rank 41 of 92	Rank 29 of 92
Price per 1M tokens
Inputwhat you send	$2above 200,000 tokens: $4	$0.15
Outputwhat the model writes	$12above 200,000 tokens: $18	$0.60
Cost per task (own tests)average of 2 original-protocol tasks both models passed (SVG short test not included)	$0.11	$0.008
Speed
Tokens per second	122	176
Popularity
Popularitysearch, usage and mentions, 0 to 100	50Rank 110 of 876	38Rank 312 of 876
Technical
Context windowhow much text the model reads at once	1.05 M tokens≈ 1,570 A4 pages	1 M tokens≈ 1,500 A4 pages
Knowledge cutoff	n/a	n/a
Released	Feb 19, 2026	Sep 10, 2026
Availability	Preview	Active
Open weights	no	yes

DeepSeek-V4.1-Flash: At peak times (Mon, Tue, Wed, Thu, Fri, 01:00-04:00 UTC; Mon, Tue, Wed, Thu, Fri, 06:00-10:00 UTC) the price is $0.30 input and $1.20 output per 1M tokens. The table shows the off-peak price.

Prices without caching or batch discounts · Page count as a rule of thumb from Artificial Analysis

Is there a better choice?

Successor of DeepSeek-V4.1-FlashDeepSeek V4.1 Flash 1017

Overall index 106.6, rank 17$0.15 input and $0.60 output per 1M tokens

Gemini 3.1 Pro Preview vs. DeepSeek V4.1 Flash 1017

Cheaper than DeepSeek-V4.1-FlashClaude Haiku 5.5

Overall index 103.4, rank 20$0.10 input and $0.50 output per 1M tokens

DeepSeek-V4.1-Flash vs. Claude Haiku 5.5

Best model in the overall indexClaude Opus 5.5

Overall index 126.1, rank 1$4 input and $20 output per 1M tokens

DeepSeek-V4.1-Flash vs. Claude Opus 5.5

Shared benchmarks

Single results from the same benchmark version, task and metric. A value is bold only with the same test setup and a clear gap.

A: Gemini 3.1 Pro PreviewB: DeepSeek-V4.1-Flash

BioMysteryBenchOverall · higher is betterTest setup not documented as identical60.0%67.8%Vals AIAs of 11/05/2026

Code MigrationOverall · higher is betterTest setup not documented as identical17.3%45.6%Vals AIAs of 10/29/2026

EMBOverall · higher is betterTest setup not documented as identical52.6%57.2%Vals AIAs of 10/20/2026

Finance Agent (v2)Overall · higher is betterTest setup not documented as identical43.0%53.5%Vals AIAs of 10/29/2026

IOIOverall · higher is betterTest setup not documented as identical51.8%40.3%Vals AIAs of 10/29/2026

Legal Research BenchOverall · All-pass · higher is betterTest setup not documented as identical20.7%41.3%Vals AIAs of 11/05/2026

LegalBenchOverall · higher is betterTest setup not documented as identical87.4%83.3%Vals AIAs of 10/13/2026

LMArena Code Arena WebDevOverall · higher is betterTest setup not documented as identical1,4471,619LMArenaAs of 10/28/2026

LMArena Text, Style ControlOverall · higher is betterTest setup not documented as identical1,4871,476LMArenaAs of 10/28/2026

LMArena Vision, Style ControlOverall · higher is betterTest setup not documented as identical1,2801,261LMArenaAs of 10/19/2026

Show 54 more benchmarks

MedCodeOverall · higher is betterTest setup not documented as identical59.1%41.2%Vals AIAs of 10/29/2026

MedScribeOverall · higher is betterTest setup not documented as identical76.1%85.5%Vals AIAs of 10/29/2026

MysteryMechanismOverall · higher is betterTest setup not documented as identical20.7%21.2%Vals AIAs of 10/29/2026

ProgramBenchOverall · higher is betterTest setup not documented as identical0.0%0.5%Vals AIAs of 11/05/2026

ProofBench v1.1Overall · higher is betterTest setup not documented as identical26%54%Vals AIAs of 10/29/2026

Public Benefits Bench v1.1Overall · higher is betterTest setup not documented as identical53.8%64.3%Vals AIAs of 10/12/2026

SAGEOverall · higher is betterTest setup not documented as identical83.2%85.4%Vals AIAs of 11/05/2026

SRE BenchOverall · higher is betterTest setup not documented as identical0.0%0.8%Vals AIAs of 11/05/2026

Tax Agent BenchOverall · higher is betterTest setup not documented as identical9.3%26.6%Vals AIAs of 11/05/2026

Terminal-Bench 4.0Overall · higher is betterTest setup not documented as identical2.5%19.7%Vals AIAs of 11/05/2026

Terminal-Bench-ScienceOverall · higher is betterTest setup not documented as identical1.4%0.0%Vals AIAs of 11/05/2026

Vals Science IndexOverall · higher is betterTest setup not documented as identical27.4%29.7%Vals AIAs of 10/25/2026

Vibe Code Bench 1-100Overall · higher is betterTest setup not documented as identical6.7%16.4%Vals AIAs of 10/29/2026

Vibe Code Bench v1.1Overall · higher is betterTest setup not documented as identical32.0%84.7%Vals AIAs of 11/05/2026

BioMysteryBenchHard · higher is betterTest setup not documented as identical27.5%21.6%Vals AIAs of 11/05/2026

Code MigrationCLI · higher is betterTest setup not documented as identical10.8%37.5%Vals AIAs of 10/29/2026

EMBComps · higher is betterTest setup not documented as identical44.5%43.9%Vals AIAs of 10/20/2026

Finance Agent (v2)Adjustments · higher is betterTest setup not documented as identical38.4%49.2%Vals AIAs of 10/29/2026

IOIIOI 2024 · higher is betterTest setup not documented as identical55.5%42.7%Vals AIAs of 10/29/2026

Legal Research BenchAdministrative / Regulatory · All-pass · higher is betterTest setup not documented as identical24.2%53.0%Vals AIAs of 11/05/2026

LegalBenchConclusion Tasks · higher is betterTest setup not documented as identical89.3%89.1%Vals AIAs of 10/13/2026

LMArena Code Arena WebDevHTML development · higher is betterTest setup not documented as identical1,4891,613LMArenaAs of 10/28/2026

LMArena Text, Style ControlBusiness, management and finance · higher is betterTest setup not documented as identical1,4771,478LMArenaAs of 10/28/2026

LMArena Vision, Style ControlChinese · higher is betterTest setup not documented as identical1,3391,311LMArenaAs of 10/19/2026

ProgramBenchAlmost Resolved · higher is betterTest setup not documented as identical1.0%16.5%Vals AIAs of 11/05/2026

SAGEComputer Science · higher is betterTest setup not documented as identical89.9%91.1%Vals AIAs of 11/05/2026

SRE BenchCapability Score · higher is betterTest setup not documented as identical1.0%14.2%Vals AIAs of 11/05/2026

Tax Agent BenchControversy & Precedence Analysis · higher is betterTest setup not documented as identical0.0%3.3%Vals AIAs of 11/05/2026

Terminal-Bench 4.0Hardware · higher is betterTest setup not documented as identical0%0%Vals AIAs of 11/05/2026

Terminal-Bench-ScienceEarth Sciences · higher is betterTest setup not documented as identical0%0%Vals AIAs of 11/05/2026

Vals Science IndexBioMysteryBench · higher is betterTest setup not documented as identical60.0%67.8%Vals AIAs of 10/25/2026

BioMysteryBenchHuman-solvable · higher is betterTest setup not documented as identical67.6%78.5%Vals AIAs of 11/05/2026

Code MigrationCOBOL · higher is betterTest setup not documented as identical36.8%70.0%Vals AIAs of 10/29/2026

EMBDataroom · higher is betterTest setup not documented as identical79.2%73.9%Vals AIAs of 10/20/2026

Finance Agent (v2)All-Pass · higher is betterTest setup not documented as identical30.9%41.1%Vals AIAs of 10/29/2026

IOIIOI 2025 · higher is betterTest setup not documented as identical46.8%35.2%Vals AIAs of 10/29/2026

Legal Research BenchAdministrative / Regulatory · Weighted partial credit (weighted pass) · higher is betterTest setup not documented as identical79.2%89.1%Vals AIAs of 11/05/2026

LegalBenchInterpretation Tasks · higher is betterTest setup not documented as identical83.8%80.4%Vals AIAs of 10/13/2026

LMArena Code Arena WebDevImage to WebDev · higher is betterTest setup not documented as identical1,4811,606LMArenaAs of 10/19/2026

LMArena Text, Style ControlChinese · higher is betterTest setup not documented as identical1,5281,523LMArenaAs of 10/28/2026

LMArena Vision, Style ControlCreative writing with images · higher is betterTest setup not documented as identical1,3011,254LMArenaAs of 10/19/2026

ProgramBenchFully Resolved · higher is betterTest setup not documented as identical0.0%0.5%Vals AIAs of 11/05/2026

SAGELinear Algebra · higher is betterTest setup not documented as identical67.2%72.7%Vals AIAs of 11/05/2026

SRE BenchFully Solved · higher is betterTest setup not documented as identical0.0%0.8%Vals AIAs of 11/05/2026

Tax Agent BenchControversy & Precedence Analysis (weighted pass) · higher is betterTest setup not documented as identical34.4%52.8%Vals AIAs of 11/05/2026

Terminal-Bench 4.0Media · higher is betterTest setup not documented as identical0.0%8.3%Vals AIAs of 11/05/2026

Terminal-Bench-ScienceEngineering Sciences · higher is betterTest setup not documented as identical0%0%Vals AIAs of 11/05/2026

Vals Science IndexMysteryMechanism · higher is betterTest setup not documented as identical20.7%21.2%Vals AIAs of 10/25/2026

Code MigrationCode Quality · higher is betterTest setup not documented as identical2.538 / 54.411 / 5Vals AIAs of 10/29/2026

EMBDCF · higher is betterTest setup not documented as identical46.4%51.0%Vals AIAs of 10/20/2026

Finance Agent (v2)Comparables · higher is betterTest setup not documented as identical36.7%46.1%Vals AIAs of 10/29/2026

IOIIOI 2026 · higher is betterTest setup not documented as identical53.2%43.0%Vals AIAs of 10/29/2026

Legal Research BenchBusiness & Commercial · All-pass · higher is betterTest setup not documented as identical23.3%39.5%Vals AIAs of 11/05/2026

LegalBenchIssue Tasks · higher is betterTest setup not documented as identical91.5%88.8%Vals AIAs of 10/13/2026

More comparisons

Compare Gemini 3.1 Pro Preview and DeepSeek-V4.1-Flash with the models people compare them with most often.

Gemini 3.1 Pro Previewvs.DeepSeek-V4.1-Flash

Choose Gemini 3.1 Pro Preview for

Choose DeepSeek-V4.1-Flash for

Where is the difference?

Strengths at a glance

Which model for which task?

See for yourself: same task, one attempt

Pelican on a bicycle

SVG short test

Gemini 3.1 Pro Preview

DeepSeek-V4.1-Flash

Gemini 3.1 Pro Preview and DeepSeek-V4.1-Flash in the overall index

Overall index (Gradually Meta-Score v4)

Popularity compared

Over time

All data side by side

Is there a better choice?

Shared benchmarks

More comparisons

Compare another matchup

Predecessor and successor

The most popular comparisons