BenchLeader

DeepSeek V4.1 Flash 1017 vs Gemini 3.1 Pro

Full profiles: DeepSeek V4.1 Flash 1017 (DeepSeek) and Gemini 3.1 Pro (Google): every benchmark, price and rank.

Updated · Data from 14 sources

As of 16 Nov 2026, DeepSeek V4.1 Flash and Gemini 3.1 Pro are neck and neck on quality (57.7 vs 60.4 on the BenchLeader Index, inside the ±4.9 margin), and DeepSeek V4.1 Flash is 8.6× cheaper at $0.525 vs $4.50 per 1M tokens blended. Choose DeepSeek V4.1 Flash for coding (63 vs 59), where it also streams faster; choose Gemini 3.1 Pro for fast first answers (15.2 s vs 39.7 s) and knowledge (66 vs 48), reasoning (65 vs 56), and maths (66 vs 61).

Key facts (as of 16 Nov 2026)

  • DeepSeek V4.1 Flash scores 57.7 on the BenchLeader Index and Gemini 3.1 Pro scores 60.4 (BenchLeader, 16 Nov 2026).
  • DeepSeek V4.1 Flash costs $0.525 per 1M tokens blended (list price, 16 Nov 2026).
  • Gemini 3.1 Pro costs $4.50 per 1M tokens blended (list price, 16 Nov 2026).
  • DeepSeek V4.1 Flash streams faster than Gemini 3.1 Pro (Artificial Analysis, 7-day median, 16 Nov 2026).
  • Of the 13 benchmarks both models have, DeepSeek V4.1 Flash leads 10 (BenchLeader, 16 Nov 2026).

Which to pick for…

The pick's numbers come first. Category scores on the BenchLeader scale (50 is the average model, 15 points one standard deviation). High confidence means a lead of half a standard deviation or more; under one point is called either.

Use casePickWhyConfidence
CodingDeepSeek V4.1 Flash 101763 vs 59 category score; Code Migration 44.2% vs 17.3%Medium
Agents & tool useEither58 vs 57 category score; Finance Agent v2 57.0% vs 43.0%Either
MathsGemini 3.1 Pro66 vs 61 category score; OTIS Mock AIME 95.6% vs 99.4%Medium
ReasoningGemini 3.1 Pro65 vs 56 category score; GPQA Diamond 94.4% vs 93.2%High
KnowledgeGemini 3.1 Pro66 vs 48 category score; SimpleQA Verified 73.5% vs 42.5%High
Lowest costDeepSeek V4.1 Flash 1017$0.525 vs $4.50 per 1M blended; agentic coding session $0.023 vs $0.168–
Fastest first answerGemini 3.1 Pro15.2 s (Gemini 3.1 Pro (high)) vs 39.7 s (DeepSeek V4.1 Flash 1017 (max)) to the first answer token–
High volume on a budgetNeither flagshipA smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs–

Key differences

  • DeepSeek V4.1 Flash 1017 (max) and Gemini 3.1 Pro (high) are neck and neck on quality (57.7 vs 60.4, within the ±4.9 margin).
  • DeepSeek V4.1 Flash 1017 (max) is stronger in agents & tools, coding.
  • Gemini 3.1 Pro (high) is stronger in knowledge, maths, reasoning.
  • DeepSeek V4.1 Flash 1017 (max) is 8.6× cheaper ($0.525 vs $4.50 per 1M blended).
  • DeepSeek V4.1 Flash 1017 (max) streams 1.6× faster (174 vs 111 tokens per second).

At a glance

SpecDeepSeek V4.1 Flash 1017Gemini 3.1 Pro
DeveloperDeepSeekGoogle
Released17 Oct 202619 Feb 2026
Context window (tokens)1,048,5761,048,576
Compared atDeepSeek V4.1 Flash 1017 (max)Gemini 3.1 Pro (high)
Open weightsYesNo
Price in / out / cached input per 1M$0.300 / $1.20 / $0.0060$2.00 / $12.00 / $0.200

Category by category

Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.

CategoryDeepSeek V4.1 Flash 1017 (max)Gemini 3.1 Pro (high)Leader
Reasoning55.764.7Gemini 3.1 Pro
Coding63.058.6DeepSeek V4.1 Flash 1017
Agents & tools57.656.7level
Maths60.965.7Gemini 3.1 Pro
Knowledge47.765.6Gemini 3.1 Pro
Multimodal63.0–not comparable
Long context66.0–not comparable
Composite55.5–not comparable

On one chart

Each domain on the index's own scale, where 50 is an average model. Click a domain to see the benchmarks behind it. Pick up to 5 models from the list.

2550 avg75ReasoningCodingAgents & toolsMathsKnowledgeDeepSeek V4.1 Flash 1017, Reasoning: 55.7DeepSeek V4.1 Flash 1017, Coding: 63.0DeepSeek V4.1 Flash 1017, Agents & tools: 57.6DeepSeek V4.1 Flash 1017, Maths: 60.9DeepSeek V4.1 Flash 1017, Knowledge: 47.7Gemini 3.1 Pro, Reasoning: 64.7Gemini 3.1 Pro, Coding: 58.6Gemini 3.1 Pro, Agents & tools: 56.7Gemini 3.1 Pro, Maths: 65.7Gemini 3.1 Pro, Knowledge: 65.6

Not drawn, measured for fewer than two of these models: Instruction following, Long context, Multimodal, Human preference.

Model, click to add or removeIndex

Where sources disagree

Across the 13 benchmarks both have, DeepSeek V4.1 Flash 1017 leads 10, Gemini 3.1 Pro leads 3.

Overall indexDeepSeek V4.1 Flash 1017 (max)Gemini 3.1 Pro (high)Leader
BenchLeader Index57.760.4neck and neck

Each publisher runs its own tasks, harnesses and effort settings, so their overall verdicts can differ; BenchLeader reconciles them on one scale. † measured at another effort setting of the same model (the publisher did not test the one compared here).

Price, speed and latency

List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page. Cost, tokens and time per benchmark task are measured: Artificial Analysis’ averages over its whole Intelligence Index run.

MeasureDeepSeek V4.1 Flash 1017 (max)Gemini 3.1 Pro (high)
Blended price per 1M tokens$0.525$4.50
Input / output per 1M$0.300 / $1.20$2.00 / $12.00
Cached input per 1M$0.0060$0.200
Price per task: chat reply$0.0005$0.0044
Price per task: summarize a 30-page report$0.0043$0.031
Price per task: code edit$0.0036$0.030
Price per task: agentic coding session$0.023$0.168
Price per task: structured extraction$0.0008$0.0064
Agentic coding session, input mostly cached$0.0096$0.087
Cost per benchmark task (AA Index)$0.602–
Output tokens per task (AA Index)139k–
Of which reasoning91k–
Time per task (AA Index)13 min–
Output speed174 tok/s111 tok/s
Time to first answer39.7 s15.2 s
Time to a complete answer, last measured 13 Nov 202691.2 s37.5 s

Each model is compared at its best-ranked reasoning-effort setting. DeepSeek V4.1 Flash 1017 (max) scores 57.7 and takes 40 s to its first answer token; DeepSeek V4.1 Flash 1017 (high) scores 51.2 in 20 s, DeepSeek V4.1 Flash 1017 (no reasoning) scores 44.9 in 1 s. Gemini 3.1 Pro (high) scores 60.4 and takes 15 s to its first answer token; Gemini 3.1 Pro scores 59.9 in 15 s. Higher effort means more thinking before the answer: better scores, longer waits and more tokens billed. How reasoning effort changes scores and cost.

Every benchmark

200 benchmarks, grouped by category; each name links to the board and its leaders. † measured at another effort setting of the same model, shown for reference and not counted as a win.

MetricPublisherDeepSeek V4.1 Flash 1017 (max)Gemini 3.1 Pro (high)
BenchLeader IndexBenchLeader57.760.4
Agents & tools scoreBenchLeader57.656.7
Coding scoreBenchLeader63.058.6
Composite scoreBenchLeader55.5–
Knowledge scoreBenchLeader47.765.6
Long context scoreBenchLeader66.0–
Maths scoreBenchLeader60.965.7
Multimodal scoreBenchLeader63.0–
Reasoning scoreBenchLeader55.764.7
Blended price $/MBenchLeader$0.525$4.50
Output speedBenchLeader174 tok/s111 tok/s
Time to first answerBenchLeader39.7 s15.2 s
Context windowBenchLeader1,048,576 tokens1,048,576 tokens
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub93.2%94.4%
Humanity's Last ExamScale AI / CAIS–46.4%†
SimpleBenchSimpleBench–79.6%†
Arena Hard PromptsArena (LMArena)–1508†
LiveBench 2.0 ReasoningLiveBench89.1%–
LiveBench 1.0 ReasoningLiveBench–84.0%
GPQA Diamond (AA)Artificial Analysis–94.1%†
Humanity's Last Exam (AA)Artificial Analysis43.3%47.0%†
HLE-Diamond (AA)Artificial Analysis25.3%31.4%†
GPQA Diamond (Vals)Vals AI–95.5%
EnigmaEvalScale AI SEAL–36.8%
ARC-AGI-1ARC Prize93.5%98.0%†
ARC-AGI-2ARC Prize78.1%77.1%†
ARC-AGI-3ARC Prize5.8%0.4%†
CritPtArtificial Analysis24.0%17.7%†
CritPt v2 hard subset (AA)Artificial Analysis25.0%–
MysteryMechanismVals AI–20.7%
HLE-Diamond (Scale)Scale AI SEAL20.5%†–
Arena ExpertArena (LMArena)–1509†
Chess PuzzlesEpoch AI Benchmarking Hub46.0%51.0%
Mystery Game PuzzlesEpoch AI Benchmarking Hub49.0%34.0%
CL-benchEpoch AI Benchmarking Hub–20.8%†
CL-bench LifeEpoch AI Benchmarking Hub–16.9%†
LMCAEpoch AI Benchmarking Hub–53.8%
DTBenchEpoch AI Benchmarking Hub–95.5%
Show all 200 benchmarks
MetricPublisherDeepSeek V4.1 Flash 1017 (max)Gemini 3.1 Pro (high)
Reasoning
ForecastBenchForecastBench–65.2%
Coding
SciCodeSciCode55.9%58.7%†
WeirdMLWeirdML–72.1%†
GSO-BenchGSO-Bench–22.6%†
Arena CodingArena (LMArena)–1520†
Arena WebDevArena (LMArena)16641446†
LiveBench 2.0 CodingLiveBench80.4%–
LiveBench 1.0 CodingLiveBench–76.5%
LiveCodeBenchVals AI–88.5%
IOIVals AI–51.8%
SWE-bench (Vals)Vals AI–78.8%
SWE-Bench ProScale AI SEAL–46.1%†
Code MigrationVals AI44.2%17.3%
Code Migration v1.1Vals AI67.5%–
ProgramBenchVals AI–0.0%
Vibe Code Bench 1-100Vals AI–6.7%
Vibe Code Bench v1.1Vals AI87.7%32.0%
Vibe Code Bench 1-100 v1.1Vals AI26.1%–
SWE-Bench Pro (private)Scale AI SEAL–32.2%†
SWE Atlas v1.0: Codebase QnAScale AI SEAL–13.5%†
SWE Atlas v1.0: RefactoringScale AI SEAL–33.8%†
SWE Atlas v1.0: Test WritingScale AI SEAL–29.8%†
MirrorCodeEpoch AI Benchmarking Hub–8.9%
DeepSWE v1.1Epoch AI Benchmarking Hub–11.7%†
CursorBenchCursor47.7%–
ALE-BenchEpoch AI Benchmarking Hub16221160.6†
AlgoTuneAlgoTune–2.0†
WeirdML v3WeirdML9.2%†–
Terminal-Bench 4.0 (AA)Artificial Analysis33.8%4.0%†
Terminal-Bench 2.1 (AA)Artificial Analysis–73.8%†
GameBench 2 OverallDuelLab GameBench 261.4†–
GameBench 2 by reasoning settingDuelLab GameBench 269.9–
Design Arena WebsiteDesign Arena1295†1226†
FrontierCode 1.1 (AA)Artificial Analysis43.7%–
FrontierCode 2.0 (AA)Artificial Analysis28.5%–
Agents & tools
Terminal-BenchTerminal-Bench38.5%80.2%†
APEX-AgentsMercor69.8%35.1%†
Arena AgentArena (LMArena)2-4.5†
LiveBench Long AgenticLiveBench38.8%–
LiveBench BrowserLiveBench67.2%–
LiveBench Agentic CodingLiveBench–44.1%
Terminal-Bench HardArtificial Analysis–53.8%†
τ²-Bench Telecom (AA)Artificial Analysis–95.6%†
Terminal-Bench 2.1 (Vals)Vals AI–70.8%
MCP AtlasScale AI SEAL–78.2%
HiL-BenchScale AI SEAL–35.3%†
τ³-bench Bankingτ²-bench–26.0%
GDPval-AA v2.1, normalised scoreArtificial Analysis62.3%14.6%†
GDPval-AA v2.1Artificial Analysis1746799†
τ³-Banking (AA)Artificial Analysis–21.4%†
ITBench SRE (AA)Artificial Analysis–30.3%†
Analyst Agent (AA)Artificial Analysis–41.3%†
APEX-Agents (AA)Artificial Analysis–32.0%†
Finance Agent v2Vals AI57.0%43.0%
Finance Agent v2.1Vals AI62.9%–
Harvey's Legal Agent Benchmark v1.1Vals AI14.2%–
Harvey's Legal Agent Benchmark v1Vals AI9.2%0.0%
Legal Research BenchVals AI45.2%20.7%
SREBenchVals AI–0.0%
Tax Agent BenchVals AI44.4%9.3%
Terminal-Bench 2.0 (Vals)Vals AI–67.4%
Terminal-Bench 4.0 (Vals)Vals AI31.8%2.5%
Terminal-Bench ScienceVals AI–1.4%
Arena SearchArena (LMArena)–1209†
Arena Agent Task CompletionArena (LMArena)2.3-11.3†
BALROGBALROG–57.0%†
DeepResearch BenchDeepResearch Bench–47.8%
PostTrainBenchEpoch AI Benchmarking Hub–18.1%†
ExploitBenchEpoch AI Benchmarking Hub–26.1%†
METR Time HorizonsMETR–77.0%†
GBAEvalEpoch AI Benchmarking Hub–0.8%†
Vending-Bench 2Andon Labs–911.2†
AutomationBenchArtificial Analysis66.6%35.4%†
GDP.pdf (AA)Artificial Analysis17.2%17.8%†
Harvey LAB-AA v1.1Harvey–0.0%†
EnterpriseOps-GymArtificial Analysis–42.2%†
AA-Briefcase v1.1Artificial Analysis1570457†
GDPxGDPx by SemiAnalysis5.4%–
Design Arena AgenticDesign Arena1282†1088†
Terminal-Bench-Science 0.1 (AA)Artificial Analysis12.9%–
CWE-Bench (AA)Artificial Analysis69.2%–
DeepSec-Bench (AA)Artificial Analysis24.9%–
CyberGym end-to-end (AA)Artificial Analysis76.3%–
MKA (AA)Artificial Analysis38.5%–
BrowseCompOpenAI–31.2%†
DeepSearchQAGoogle–60.2%†
FACTS SearchGoogle–90.2%†
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub71.2%59.6%†
FrontierMath Tier 4Epoch AI Benchmarking Hub46.3%26.8%†
OTIS Mock AIMEEpoch AI Benchmarking Hub99.4%95.6%
ProofBenchVals AI–26.0%†
LiveBench MathematicsLiveBench–91.0%
AIME (Vals)Vals AI–98.1%
AIME 2026MathArena–98.3%
HMMT February 2026MathArena–94.7%
USAMO 2026MathArena–74.4%
MathArena ApexMathArena–60.9%
ArXivMathMathArena42.5%–
BrokenArXivMathArena48.7%–
Arena MathsArena (LMArena)–1488†
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub42.5%73.5%
LiveBench 2.0 Data AnalysisLiveBench79.5%–
LiveBench ResearchLiveBench78.6%–
LiveBench 1.0 Data AnalysisLiveBench–78.5%
AA-Omniscience v1.1Artificial Analysis0.9–
MMLU-ProVals AI–91.0%
LegalBenchVals AI–87.4%
CorpFinVals AI–64.5%
TaxEvalVals AI–72.9%
MedQAVals AI–96.4%
PRBench FinanceScale AI SEAL47.1%†41.9%†
PRBench LegalScale AI SEAL47.3%†44.0%†
MultiNRCScale AI SEAL–64.7%†
BioMysteryBenchVals AI–60.0%
CaseLaw v2Vals AI–64.8%
Excel Modeling BenchmarkVals AI61.8%52.6%
Excel Modeling Benchmark v1.1Vals AI69.2%–
MedCodeVals AI–59.1%
MedScribeVals AI–76.1%
Public Benefits Bench v1.1Vals AI–53.8%
Access to Justice BenchmarkVals AI73.1%–
Financial Coaching BenchVals AI–66.7%
DrugDiscoveryBenchScale AI SEAL60.6%†41.9%
EBR-benchEpoch AI Benchmarking Hub–14.3%†
GDP.pdfEpoch AI Benchmarking Hub–17.0%
AA-Omniscience v1.1: accuracyArtificial Analysis39.6%–
AA-Omniscience v1.1: non-hallucinationArtificial Analysis35.9%–
Global-MMLU-Lite (AA)Artificial Analysis–93.2%†
FACTS ParametricGoogle–79.4%†
Instruction following
LiveBench LanguageLiveBench–85.4%
LiveBench Instruction FollowingLiveBench–79.1%
IFBenchArtificial Analysis–77.1%†
MultiChallengeScale AI SEAL74.9%†71.4%†
TutorBenchScale AI SEAL–53.0%†
Arena Instruction FollowingArena (LMArena)–1480†
Human preference
Arena TextArena (LMArena)–1487†
EQ-Bench 4EQ-Bench–1132†
EQ-Bench Creative Writing v3EQ-Bench–1491†
Arena Creative WritingArena (LMArena)–1482†
Arena Multi-turnArena (LMArena)–1495†
Arena Longer QueriesArena (LMArena)–1499†
Multimodal
Arena VisionArena (LMArena)–1296†
LiveBench DesignLiveBench79.8%–
MMMU-ProArtificial Analysis80.6%82.4%†
MMMU-Pro (Vals)Vals AI–88.2%
MortgageTaxVals AI–69.4%
SAGEVals AI–83.2%
Humanity's Sixth SenseScale AI SEAL30.1%–
VTBScale AI SEAL–29.0%†
GeoBenchGeoBench–84.0%†
Blueprint-Bench 2Epoch AI Benchmarking Hub–26.5%†
Furniture AssemblyEpoch AI Benchmarking Hub–26.7%
VIALS-AAHandshake AI40.4%–
FACTS MultimodalGoogle–50.4%†
Long context
AA-LCR v1.1Artificial Analysis82.7%82.0%†
Arena DocumentArena (LMArena)–1459†
MLCRArtificial Analysis26.7%11.1%†
FACTS GroundingGoogle–56.6%†
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub159.4†156.2†
LiveBench 2.0LiveBench70.3%–
LiveBench 1.0LiveBench–77.0%
AA Intelligence Index v5.0.1Artificial Analysis32.6–
AA Intelligence Index v4.3.2Artificial Analysis45.129.7†
AA Progress Index v1Artificial Analysis323.7296.1†
Vals IndexVals AI61.3–
Vals Finance IndexVals AI60.9%–
Vals Science IndexVals AI–27.4%
Vals Multimodal IndexVals AI–56.1%
Vals Router IndexVals AI61.3%†–
Vals Legal IndexVals AI71.9%–
AA Cyber IndexArtificial Analysis56.8–
Safety & honesty
FORTRESSScale AI SEAL27.2%†29.8%†
MASKScale AI SEAL–42.4%†
DistressBenchScale AI SEAL–68.8%†
Metadata
AA Openness IndexArtificial Analysis44.4–

Data as of 2026-11-16. Best configuration of each model; every score links to its source on the model pages.

DuelLab GameBench 2, release gb2-20261111-7b598bb13de4 (https://benchmarks.duellab.org/releases/gb2-20261111-7b598bb13de4/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.

What changed

  • DeepSeek V4.1 Flash 1017 (max): 8 new results (LiveBench 2.0, LiveBench Long Agentic, LiveBench 2.0 Reasoning…).
  • DeepSeek V4.1 Flash 1017 (max): BenchLeader Index 58.5 → 58.0.
  • DeepSeek V4.1 Flash 1017 (max): 1 new result (Terminal-Bench); 1 score revision.
  • DeepSeek V4.1 Flash 1017 (max): BenchLeader Index 59.0 → 58.5.
  • Gemini 3.1 Pro (high): 4 new results (AIME 2026, HMMT February 2026, USAMO 2026…).
  • DeepSeek V4.1 Flash 1017 (max): 1 score revision.
  • DeepSeek V4.1 Flash 1017 (max): 1 score revision.
  • DeepSeek V4.1 Flash 1017 (max): 3 score revisions.
  • DeepSeek V4.1 Flash 1017 (max): 1 new result (ARC-AGI-3); 2 score revisions.
  • DeepSeek V4.1 Flash 1017 (max): 1 score revision.

Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs

DeepSeek V4.1 Flash 1017 vs Gemini 3.1 Pro: questions

Is DeepSeek V4.1 Flash 1017 better than Gemini 3.1 Pro?
DeepSeek V4.1 Flash 1017 (max) and Gemini 3.1 Pro (high) are neck and neck on quality (57.7 vs 60.4, within the ±4.9 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 16 Nov 2026, so check the category scores for your use.
DeepSeek V4.1 Flash 1017 vs Gemini 3.1 Pro: which is cheaper, and will that change?
DeepSeek V4.1 Flash 1017 is cheaper today: $0.525 against $4.50 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
Is DeepSeek V4.1 Flash 1017 or Gemini 3.1 Pro better for coding?
DeepSeek V4.1 Flash 1017 scores higher in coding (63 vs 59 on the category index, where 50 is average).
Which is faster, DeepSeek V4.1 Flash 1017 or Gemini 3.1 Pro?
DeepSeek V4.1 Flash 1017 streams faster (174 vs 111 output tokens per second, Artificial Analysis 7-day median).
Is DeepSeek V4.1 Flash the same as DeepSeek V4.1 Flash 1017?
DeepSeek V4.1 Flash 1017 is the checkpoint compared here (BenchLeader Index 57.7); people searching for DeepSeek V4.1 Flash land on this page because it replaced that earlier build.
More questions (5)
Is DeepSeek V4.1 Flash 1017 or Gemini 3.1 Pro better for agents and tool use?
They are level in agentic tasks (58 vs 57 on the category index).
Which is better for long documents?
Both accept 1,048,576 tokens of context.
Which has lower latency?
Gemini 3.1 Pro answers first: 15.2 s against 39.7 s to the first answer token, as Artificial Analysis measures it. Reasoning effort drives most of this: a higher setting thinks longer before it answers.
Do independent benchmarks agree on DeepSeek V4.1 Flash 1017 vs Gemini 3.1 Pro?
Across the 13 benchmarks both have, DeepSeek V4.1 Flash 1017 leads 10, Gemini 3.1 Pro leads 3.
When was DeepSeek V4.1 Flash 1017 released?
DeepSeek V4.1 Flash 1017 was released on 17 Oct 2026.