Gemini 3.1 Pro Previewvs.DeepSeek V4.1 Flash 1017

Which one is better?

DeepSeek V4.1 Flash 1017 is clearly ahead in the overall index and is cheaper too: Gemini 3.1 Pro Preview costs 13.3 times (input) and 20 times (output) as much per token.

IntelligenceOverall index, 0 to 100 · higher is better

Gemini 3.1 Pro Preview

Rank 36 to 44

DeepSeek V4.1 Flash 1017

Rank 14 to 22

SpeedTokens per second · higher is better

145

Gemini 3.1 Pro Preview

fairly fast

178

DeepSeek V4.1 Flash 1017

fast

PriceDollars per 1M tokens · lower is better

$4.50

Gemini 3.1 Pro Preview

expensive

$0.26

DeepSeek V4.1 Flash 1017

cheap

Gemini 3.1 Pro Preview

Subscription: No Google subscription includes this model. Through other providers from ≈ €6.36 a month (ElevenLabs Starter, with limited credits). 26 plans in Germany.
Via API: $2 input, $12 output per 1M tokens, unchanged since our price tracking began on Aug 14, 2026.

DeepSeek V4.1 Flash 1017

Subscription: No DeepSeek subscription includes this model. Through other providers from ≈ €6.36 a month (Token Plan Individual Lite, with limited credits). 7 plans in Germany.
Via API: $0.15 input, $0.60 output per 1M tokens according to the provider; the price index has listed it at $0.30 and $1 since Oct 26, 2026.

Overall index with data through Oct 29, 2026 · Prices from Oct 29, 2026 · Speed: measured by Artificial Analysis · Subscriptions as of Oct 24, 2026 · Subscription prices incl. VAT

Where is the difference?

Values from 0 to 100 per task area. We only name a winner when the gap is large enough and there are enough tests.

Gemini 3.1 Pro Preview ahead: 1DeepSeek V4.1 Flash 1017 ahead: 4level: 1too little data: 4of 10 areas

AreaGemini 3.1 Pro PreviewDeepSeek V4.1 Flash 10170 to 100, higher is betterResult

Areas in the overall index

CodingDeepSeek V4.1 Flash 1017 ahead1172

Reasoninglevel5045

Mathematicsonly Gemini 3.1 Pro Preview has a score16not enough data

KnowledgeGemini 3.1 Pro Preview ahead4733

AgentsDeepSeek V4.1 Flash 1017 ahead2475

Generalonly Gemini 3.1 Pro Preview has a score29not enough data

Multimodaltoo little data4255

Specialist fields

FinanceDeepSeek V4.1 Flash 1017 ahead1069

LegalDeepSeek V4.1 Flash 1017 ahead1676

Healthonly Gemini 3.1 Pro Preview has a score33not enough data

Show 3 more areas that do not count toward the overall index

AreaGemini 3.1 Pro PreviewDeepSeek V4.1 Flash 10170 to 100, higher is betterResult

Searchonly Gemini 3.1 Pro Preview has a score32not enough data

Productivityonly Gemini 3.1 Pro Preview has a score30not enough data

Educationonly Gemini 3.1 Pro Preview has a score12not enough data

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

A community classic for free-form SVG drawing.

Gemini 3.1 Pro Preview

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

Reasoning (requested): high

Current attempt

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

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

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 1017

Not tested yet

Prompt and test conditions

Generate an SVG of a pelican riding a bicycle

Token limit including reasoning: 8,192 tokens

Task origin (Simon Willison)

SVG short test

There is no shared test round for this selection. Results from different rounds are not mixed.

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 1017
Intelligence
Overall index0 to 100, higher is better	26	64
Rank	Rank 36 to 44 of 95	Rank 14 to 22 of 95
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
Speed
Tokens per second	145	178
Technical
Context windowhow much text the model reads at once	1.05 M tokens≈ 1,570 A4 pages	1.05 M tokens≈ 1,570 A4 pages
Knowledge cutoff	n/a	n/a
Released	Feb 19, 2026	Oct 17, 2026
Availability	Preview	Open Source
Open weights	no	yes

DeepSeek V4.1 Flash 1017: 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.
DeepSeek V4.1 Flash 1017, Time-based API pricing: Off peak: $0.15 input, $0.60 output, and $0.003 cached input. Weekdays UTC 01:00-04:00 and 06:00-10:00: $0.30, $1.20, and $0.006. Chinese public holidays are excluded. Source

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

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 1017

Code MigrationOverall · higher is betterTest setup not documented as identical17.3%44.2%Vals AIAs of 10/18/2026

EMBOverall · higher is betterTest setup not documented as identical52.6%61.8%Vals AIAs of 10/20/2026

Finance Agent (v2)Overall · higher is betterTest setup not documented as identical43%57%Vals AIAs of 10/20/2026

Legal Research BenchOverall · All-pass · higher is betterTest setup not documented as identical20.7%45.2%Vals AIAs of 10/26/2026

LMArena Code Arena WebDevOverall · higher is betterTest setup not documented as identical1,4471,666LMArenaAs of 10/27/2026

Tax Agent BenchOverall · higher is betterTest setup not documented as identical9.3%44.4%Vals AIAs of 10/26/2026

Terminal-Bench 4.0Overall · higher is betterTest setup not documented as identical2.5%31.8%Vals AIAs of 10/26/2026

Vibe Code Bench v1.1Overall · higher is betterTest setup not documented as identical32%87.7%Vals AIAs of 10/26/2026

Code MigrationCLI · higher is betterTest setup not documented as identical10.8%35.6%Vals AIAs of 10/18/2026

EMBComps · higher is betterTest setup not documented as identical44.5%49.8%Vals AIAs of 10/20/2026

Show 54 more benchmarks

Finance Agent (v2)Adjustments · higher is betterTest setup not documented as identical38.4%58.4%Vals AIAs of 10/20/2026

Legal Research BenchAdministrative / Regulatory · All-pass · higher is betterTest setup not documented as identical24.2%59.1%Vals AIAs of 10/26/2026

LMArena Code Arena WebDevHTML development · higher is betterTest setup not documented as identical1,4891,705LMArenaAs of 10/27/2026

Tax Agent BenchControversy & Precedence Analysis · higher is betterTest setup not documented as identical0%23.3%Vals AIAs of 10/26/2026

Terminal-Bench 4.0Hardware · higher is betterTest setup not documented as identical0%6.7%Vals AIAs of 10/26/2026

Code MigrationCOBOL · higher is betterTest setup not documented as identical36.8%70%Vals AIAs of 10/18/2026

EMBDataroom · higher is betterTest setup not documented as identical79.2%82.8%Vals AIAs of 10/20/2026

Finance Agent (v2)All-Pass · higher is betterTest setup not documented as identical30.9%45.2%Vals AIAs of 10/20/2026

Legal Research BenchAdministrative / Regulatory · Weighted partial credit · higher is betterTest setup not documented as identical79.2%91.3%Vals AIAs of 10/26/2026

LMArena Code Arena WebDevReact development · higher is betterTest setup not documented as identical1,4381,672LMArenaAs of 10/27/2026

Tax Agent BenchControversy & Precedence Analysis · higher is betterTest setup not documented as identical34.4%73.2%Vals AIAs of 10/26/2026

Terminal-Bench 4.0Media · higher is betterTest setup not documented as identical0%25%Vals AIAs of 10/26/2026

Code MigrationCode Quality · higher is betterTest setup not documented as identical2.538 / 53.992 / 5Vals AIAs of 10/18/2026

EMBDCF · higher is betterTest setup not documented as identical46.4%53%Vals AIAs of 10/20/2026

Finance Agent (v2)Comparables · higher is betterTest setup not documented as identical36.7%47.5%Vals AIAs of 10/20/2026

Legal Research BenchBusiness & Commercial · All-pass · higher is betterTest setup not documented as identical23.3%44.2%Vals AIAs of 10/26/2026

LMArena Code Arena WebDevWeb development · higher is betterTest setup not documented as identical1,4471,666LMArenaAs of 10/27/2026

Tax Agent BenchCurrent & Temporal Analysis · higher is betterTest setup not documented as identical9.8%56.4%Vals AIAs of 10/26/2026

Terminal-Bench 4.0ML · higher is betterTest setup not documented as identical0%42.4%Vals AIAs of 10/26/2026

EMBLBO · higher is betterTest setup not documented as identical38%57.8%Vals AIAs of 10/20/2026

Finance Agent (v2)Disclosure Analysis · higher is betterTest setup not documented as identical46.2%66.3%Vals AIAs of 10/20/2026

Legal Research BenchBusiness & Commercial · Weighted partial credit · higher is betterTest setup not documented as identical80.8%87.8%Vals AIAs of 10/26/2026

Tax Agent BenchCurrent & Temporal Analysis · higher is betterTest setup not documented as identical36.5%75%Vals AIAs of 10/26/2026

Terminal-Bench 4.0Operations · higher is betterTest setup not documented as identical7.4%18.5%Vals AIAs of 10/26/2026

EMBM&A · higher is betterTest setup not documented as identical58.8%62.3%Vals AIAs of 10/20/2026

Finance Agent (v2)Earnings Analysis · higher is betterTest setup not documented as identical64.7%75.3%Vals AIAs of 10/20/2026

Legal Research BenchCivil Litigation · All-pass · higher is betterTest setup not documented as identical11.1%37%Vals AIAs of 10/26/2026

Tax Agent BenchFact-Pattern Analysis · higher is betterTest setup not documented as identical7%55.8%Vals AIAs of 10/26/2026

Terminal-Bench 4.0Science · higher is betterTest setup not documented as identical2.4%38.1%Vals AIAs of 10/26/2026

EMBMisc · higher is betterTest setup not documented as identical43.4%60.9%Vals AIAs of 10/20/2026

Finance Agent (v2)Financial Modeling · higher is betterTest setup not documented as identical15.1%22.8%Vals AIAs of 10/20/2026

Legal Research BenchCivil Litigation · Weighted partial credit · higher is betterTest setup not documented as identical82.7%84.9%Vals AIAs of 10/26/2026

Tax Agent BenchFact-Pattern Analysis · higher is betterTest setup not documented as identical32.9%72.5%Vals AIAs of 10/26/2026

Terminal-Bench 4.0Security · higher is betterTest setup not documented as identical0%26.7%Vals AIAs of 10/26/2026

EMBOperating · higher is betterTest setup not documented as identical58.1%66.1%Vals AIAs of 10/20/2026

Finance Agent (v2)General Qualitative Analysis · higher is betterTest setup not documented as identical45.2%74.1%Vals AIAs of 10/20/2026

Legal Research BenchConstitutional / Civil Rights · All-pass · higher is betterTest setup not documented as identical21.1%42.1%Vals AIAs of 10/26/2026

Tax Agent BenchForms & Filings · higher is betterTest setup not documented as identical3.3%33.3%Vals AIAs of 10/26/2026

Terminal-Bench 4.0Software · higher is betterTest setup not documented as identical3.7%37%Vals AIAs of 10/26/2026

EMBScratch · higher is betterTest setup not documented as identical57.7%68.5%Vals AIAs of 10/20/2026

Finance Agent (v2)General Quantitative Analysis · higher is betterTest setup not documented as identical67%79.2%Vals AIAs of 10/20/2026

Legal Research BenchConstitutional / Civil Rights · Weighted partial credit · higher is betterTest setup not documented as identical79.6%86.6%Vals AIAs of 10/26/2026

Tax Agent BenchForms & Filings · higher is betterTest setup not documented as identical30.7%79.4%Vals AIAs of 10/26/2026

EMBTemplate · higher is betterTest setup not documented as identical47.2%54.7%Vals AIAs of 10/20/2026

Finance Agent (v2)Market Analysis · higher is betterTest setup not documented as identical52.8%62.7%Vals AIAs of 10/20/2026

Legal Research BenchCriminal · All-pass · higher is betterTest setup not documented as identical19.3%35.1%Vals AIAs of 10/26/2026

Tax Agent BenchNumbers & Calculations · higher is betterTest setup not documented as identical23.3%45.7%Vals AIAs of 10/26/2026

Finance Agent (v2)Precedents · higher is betterTest setup not documented as identical20.8%26.5%Vals AIAs of 10/20/2026

Legal Research BenchCriminal · Weighted partial credit · higher is betterTest setup not documented as identical76.2%83.2%Vals AIAs of 10/26/2026

Tax Agent BenchNumbers & Calculations · higher is betterTest setup not documented as identical66.5%79.5%Vals AIAs of 10/26/2026

Legal Research BenchFamily · All-pass · higher is betterTest setup not documented as identical4.5%27.3%Vals AIAs of 10/26/2026

Tax Agent BenchRule & Source Lookup · higher is betterTest setup not documented as identical13.3%46.7%Vals AIAs of 10/26/2026

Legal Research BenchFamily · Weighted partial credit · higher is betterTest setup not documented as identical58.2%74.2%Vals AIAs of 10/26/2026

Tax Agent BenchRule & Source Lookup · higher is betterTest setup not documented as identical28.3%70.8%Vals AIAs of 10/26/2026

Compare another matchup

First modelSecond model

More matchups

Compare Gemini 3.1 Pro Preview and DeepSeek V4.1 Flash 1017 with other leading models.

View the complete LLM comparison

Gemini 3.1 Pro Previewvs.DeepSeek V4.1 Flash 1017

Gemini 3.1 Pro Preview

DeepSeek V4.1 Flash 1017

Where is the difference?

See for yourself: same task, one attempt

Pelican on a bicycle

All data side by side

Shared benchmarks

Compare another matchup

Related LLM comparisons