BenchLeader
Menu
Best LLMsChoose a modelLeaderboardsModelsBenchmarksProvidersReleases9

Gemini 3.5 Flash vs GPT-5.4 Pro

Full profiles: Gemini 3.5 Flash (Google) and GPT-5.4 Pro (OpenAI): every benchmark, price and rank.

Updated · Data from 9 sources

As of 25 Oct 2026, Gemini 3.5 Flash (high) and GPT-5.4 Pro are neck and neck on quality (63.1 vs 63.3 on the BenchLeader Index, inside the ±4.4 margin of error), and Gemini 3.5 Flash (high) costs 20× less, which makes it the better buy for most workloads. Pick Gemini 3.5 Flash for instruction following (74 vs 64); pick GPT-5.4 Pro for reasoning (73 vs 61).

Gemini 3.5 Flash benchmarks high Google GPT-5.4 Pro benchmarks OpenAIOpen in the comparison tool

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
ReasoningGPT-5.4 Pro73 vs 61 category scoreHigh
KnowledgeGPT-5.4 Pro74 vs 65 category scoreHigh
Images and other mediaGPT-5.4 Pro70 vs 67 category scoreMedium
Following instructionsGemini 3.5 Flash74 vs 64 category scoreHigh
Chat quality (human preference)–Only Gemini 3.5 Flash has an Arena rating (67 category score)Not comparable
Lowest costGemini 3.5 Flash$3.38 vs $67.50 per 1M blended; agentic coding session $0.126 vs $2.52–
Fastest first answerGPT-5.4 Pro6.1 s (GPT-5.4 Pro) vs 14.2 s (Gemini 3.5 Flash (high)) 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

  • Gemini 3.5 Flash (high) and GPT-5.4 Pro are neck and neck on quality (63.1 vs 63.3, within the ±4.4 margin).
  • Gemini 3.5 Flash (high) is stronger in agents & tools, coding, composite, human preference, instruction following, long context, maths.
  • GPT-5.4 Pro is stronger in knowledge, multimodal, reasoning.
  • Gemini 3.5 Flash (high) is 20× cheaper ($3.38 vs $67.50 per 1M blended).
  • Gemini 3.5 Flash (high) streams 31.7× faster (222 vs 7 tokens per second).

At a glance

SpecGemini 3.5 FlashGPT-5.4 Pro
DeveloperGoogleOpenAI
Released19 May 20264 Mar 2026
Context window (tokens)1,048,5761,050,000
Compared atGemini 3.5 Flash (high)GPT-5.4 Pro
Open weightsNoNo
Price in / out / cached input per 1M$1.50 / $9.00 / $0.150$30.00 / $180.00 / –

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.

CategoryGemini 3.5 Flash (high)GPT-5.4 ProLeader
Reasoning61.372.5GPT-5.4 Pro
Coding59.8–not comparable
Agents & tools63.4–not comparable
Maths58.3–not comparable
Knowledge65.273.6GPT-5.4 Pro
Instruction following74.363.8Gemini 3.5 Flash
Human preference66.8–not comparable
Multimodal66.669.8GPT-5.4 Pro
Long context61.6–not comparable
Composite65.9–not comparable

Where sources disagree

Across the 0 benchmarks both have, Gemini 3.5 Flash leads 0, GPT-5.4 Pro leads 0.

Overall indexGemini 3.5 Flash (high)GPT-5.4 ProLeader
BenchLeader Index63.163.3neck 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; suite cost is what Artificial Analysis paid to run its whole Intelligence Index.

MeasureGemini 3.5 Flash (high)GPT-5.4 Pro
Blended price per 1M tokens$3.38$67.50
Input / output per 1M$1.50 / $9.00$30.00 / $180.00
Cached input per 1M$0.150–
Price per task: chat reply$0.0033$0.066
Price per task: summarize a 30-page report$0.023$0.468
Price per task: code edit$0.022$0.450
Price per task: agentic coding session$0.126$2.52
Price per task: structured extraction$0.0048$0.096
Cost to run a full benchmark suite$3.69–
Output speed222 tok/s7 tok/s
Time to first answer14.2 s6.1 s
Time to a complete answer16.4 s–

Each model is compared at its best-ranked reasoning-effort setting. Gemini 3.5 Flash (high) scores 63.1 and takes 14 s to its first answer token; Gemini 3.5 Flash (medium) scores 62.8 in 12 s, Gemini 3.5 Flash (minimal) scores 56.4 in 1 s. GPT-5.4 Pro scores 63.3 and takes 6 s to its first answer token; GPT-5.4 Pro (xhigh) scores 61.9 in 6 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

113 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.

MetricPublisherGemini 3.5 Flash (high)GPT-5.4 Pro
BenchLeader IndexBenchLeader63.163.3
Agents & tools scoreBenchLeader63.4–
Coding scoreBenchLeader59.8–
Composite scoreBenchLeader65.9–
Human preference scoreBenchLeader66.8–
Instruction following scoreBenchLeader74.363.8
Knowledge scoreBenchLeader65.273.6
Long context scoreBenchLeader61.6–
Maths scoreBenchLeader58.3–
Multimodal scoreBenchLeader66.669.8
Reasoning scoreBenchLeader61.372.5
Blended price $/MBenchLeader$3.38$67.50
Output speedBenchLeader222 tok/s7 tok/s
Time to first answerBenchLeader14.2 s6.1 s
Context windowBenchLeader1,048,576 tokens1,050,000 tokens
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub92.8%94.6%†
Humanity's Last ExamScale AI / CAIS–44.3%
SimpleBenchSimpleBench76.7%†74.1%
Arena Hard PromptsArena (LMArena)1495–
LiveBench 1.0 ReasoningLiveBench82.0%–
GPQA Diamond (AA)Artificial Analysis92.2%–
Humanity's Last Exam (AA)Artificial Analysis42.7%–
GPQA Diamond (Vals)Vals AI92.7%–
EnigmaEvalScale AI SEAL25.4%–
ARC-AGI-1ARC Prize92.5%94.5%†
ARC-AGI-2ARC Prize72.1%83.3%†
CritPtArtificial Analysis13.1%30.0%†
Chess PuzzlesEpoch AI Benchmarking Hub45.0%59.0%†
Mystery Game PuzzlesEpoch AI Benchmarking Hub32.0%–
LMCAEpoch AI Benchmarking Hub47.1%–
DTBenchEpoch AI Benchmarking Hub94.7%–
ForecastBenchForecastBench64.3%†–
Coding
SWE-bench Verified (Epoch)Epoch AI Benchmarking Hub79.3%–
SciCodeSciCode53.9%–
WeirdMLWeirdML62.6%57.4%†
Arena CodingArena (LMArena)1508–
Arena WebDevArena (LMArena)1499–
LiveBench 1.0 CodingLiveBench78.2%–
LiveCodeBenchVals AI87.6%–
SWE-bench (Vals)Vals AI78.8%–
Show all 113 benchmarks
MetricPublisherGemini 3.5 Flash (high)GPT-5.4 Pro
Coding
Code MigrationVals AI26.8%–
ProgramBenchVals AI0.0%–
Vibe Code Bench v1.1Vals AI48.7%–
Surface Evolver BenchEpoch AI Benchmarking Hub58.1%†–
DeepSWE v1.1Epoch AI Benchmarking Hub36.1%–
ALE-BenchEpoch AI Benchmarking Hub911.0–
Terminal-Bench 4.0 (AA)Artificial Analysis6.6%–
Terminal-Bench 2.1 (AA)Artificial Analysis78.7%–
Agents & tools
APEX-AgentsMercor27.5%†–
LiveBench Agentic CodingLiveBench49.0%–
Terminal-Bench HardArtificial Analysis40.9%–
τ²-Bench Telecom (AA)Artificial Analysis95.3%–
Terminal-Bench 2.1 (Vals)Vals AI74.2%–
MCP AtlasScale AI SEAL83.6%–
HiL-BenchScale AI SEAL27.7%†–
τ³-Banking (AA)Artificial Analysis32.2%–
ITBench SRE (AA)Artificial Analysis40.4%–
Analyst Agent (AA)Artificial Analysis45.0%–
APEX-Agents (AA)Artificial Analysis47.0%–
Finance Agent v2Vals AI57.9%–
Harvey's Legal Agent Benchmark v1Vals AI2.5%–
Legal Research BenchVals AI30.8%–
SkillsBenchVals AI52.7%–
Tax Agent BenchVals AI21.6%–
Terminal-Bench 2.0 (Vals)Vals AI67.4%–
Terminal-Bench 4.0 (Vals)Vals AI6.1%–
Terminal-Bench ScienceVals AI5.7%–
Web Search IndexVals AI37.0%–
GBAEvalEpoch AI Benchmarking Hub6.7%†–
Vending-Bench 2Andon Labs5396.4†–
AutomationBenchArtificial Analysis42.1%–
GDP.pdf (AA)Artificial Analysis19.8%–
EnterpriseOps-GymArtificial Analysis50.1%–
AA-Briefcase v1.1Artificial Analysis874–
FACTS SearchGoogle64.3%†–
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub62.8%82.5%†
FrontierMath Tier 4Epoch AI Benchmarking Hub26.8%58.5%†
OTIS Mock AIMEEpoch AI Benchmarking Hub95.6%–
ProofBenchVals AI31.0%–
LiveBench MathematicsLiveBench88.2%–
AIME 2026MathArena95.0%†–
HMMT February 2026MathArena95.5%†–
MathArena ApexMathArena32.3%†69.8%†
Arena MathsArena (LMArena)1497–
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub66.2%46.3%†
LiveBench 1.0 Data AnalysisLiveBench64.9%–
AA-OmniscienceArtificial Analysis21.2–
MMLU-ProVals AI89.5%–
LegalBenchVals AI83.6%–
CorpFinVals AI64.7%–
TaxEvalVals AI74.4%–
MultiNRCScale AI SEAL–62.3%
Excel Modeling BenchmarkVals AI63.5%–
MedCodeVals AI55.8%–
MedScribeVals AI76.6%–
Public Benefits Bench v1.1Vals AI59.5%–
DrugDiscoveryBenchScale AI SEAL50.0%–
EBR-benchEpoch AI Benchmarking Hub4.8%–
GDP.pdfEpoch AI Benchmarking Hub14.0%†–
AA-Omniscience: accuracyArtificial Analysis51.4%–
AA-Omniscience: non-hallucinationArtificial Analysis37.8%–
Instruction following
LiveBench LanguageLiveBench84.6%–
LiveBench Instruction FollowingLiveBench75.6%–
IFBenchArtificial Analysis76.3%–
MultiChallengeScale AI SEAL–69.2%
TutorBenchScale AI SEAL–56.6%
Arena Instruction FollowingArena (LMArena)1465–
Human preference
Arena TextArena (LMArena)1477–
EQ-Bench 4EQ-Bench1081†–
Arena Creative WritingArena (LMArena)1464–
Arena Multi-turnArena (LMArena)1480–
Arena Longer QueriesArena (LMArena)1482–
Multimodal
Arena VisionArena (LMArena)1308–
MMMU-ProArtificial Analysis84.3%–
VISTAScale AI SEAL–53.9%
MMMU-Pro (Vals)Vals AI88.3%–
MortgageTaxVals AI68.1%–
SAGEVals AI49.9%–
Blueprint-Bench 2Epoch AI Benchmarking Hub33.6%†–
Long context
AA-LCRArtificial Analysis73.3%–
Arena DocumentArena (LMArena)1453–
MLCRArtificial Analysis10.0%–
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub154.9†159.6
LiveBench 1.0LiveBench74.6%–
AA Intelligence Index v4.3.2Artificial Analysis32.6–
Vals Multimodal IndexVals AI62.9%–
Safety & honesty
FORTRESSScale AI SEAL–14.8%
MASKScale AI SEAL–91.7%

Data as of 2026-10-25. Best configuration of each model; every score links to its source on the model pages.

What changed

  • Gemini 3.5 Flash (high): 1 new result (Harvey's Legal Agent Benchmark v1).
  • Gemini 3.5 Flash (high): 1 score revision.
  • GPT-5.4 Pro: 1 score revision.
  • Gemini 3.5 Flash (high): BenchLeader Index 62.5 → 63.2.
  • Gemini 3.5 Flash (high): 3 new results (GDPval-AA v2.1, AA-Omniscience: accuracy, AA-Omniscience: non-hallucination).
  • Gemini 3.5 Flash (high): 1 score revision.
  • Gemini 3.5 Flash (high): 2 score revisions.
  • Gemini 3.5 Flash (high): BenchLeader Index 63.0 → 62.5.
  • Gemini 3.5 Flash (high): 1 new result (DrugDiscoveryBench).
  • GPT-5.4 Pro: BenchLeader Index 62.1 → 63.6.

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

Gemini 3.5 Flash vs GPT-5.4 Pro: questions

Is Gemini 3.5 Flash better than GPT-5.4 Pro?
Gemini 3.5 Flash (high) and GPT-5.4 Pro are neck and neck on quality (63.1 vs 63.3, within the ±4.4 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 2026-10-25, so check the category scores for your use.
Which is cheaper, Gemini 3.5 Flash or GPT-5.4 Pro, and will that change?
Gemini 3.5 Flash is cheaper today: $3.38 against $67.50 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
Which is better for long documents?
Both take about 1.1M tokens: Gemini 3.5 Flash 1,048,576, GPT-5.4 Pro 1,050,000.
Which has lower latency?
GPT-5.4 Pro answers first: 6.1 s against 14.2 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.
Which is faster, Gemini 3.5 Flash or GPT-5.4 Pro?
Gemini 3.5 Flash streams faster: 222 against 7 output tokens per second.
When was Gemini 3.5 Flash released?
Gemini 3.5 Flash was released on 19 May 2026.