LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

Claude Haiku 4.5 vs Gemini 3.6 Flash

Updated October 21, 2026. 74 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace Claude Haiku 4.5 Replace Gemini 3.6 Flash
Claude Haiku 4.5Anthropic
Coding
49.8
Overall
16.9
Output $/M
$5.00
1
Gemini 3.6 FlashGoogle
Coding
48.3
Overall
34.0
Output $/M
$3.75
72
vs boards won of 74 shared, 1 tied

Decision reading

Claude Haiku 4.5 and Gemini 3.6 Flash are level for coding, 49.8 against 48.3 on the Vibe Coding Index, so price and context decide.

Pick Gemini 3.6 Flash when output tokens cost 1.3x less, or when the context window is larger, 1.05M against 200K tokens. Pick Claude Haiku 4.5 when a solved task costs $0.56 against $1.49, or when the benchmark that matters is MedScribe (Vals AI), where it leads.

Model AClaude Haiku 4.5Anthropic49.8/1003 independent benchmarks$0.56 per solved taskModel BGemini 3.6 FlashGoogle48.3/1001 independent benchmark$1.49 per solved task
Shared results
74
Claude Haiku 4.5 only
20
Gemini 3.6 Flash only
33
Like-for-like categories
8 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 Haiku 4.5 49.8 from 3 benchmarks, Gemini 3.6 Flash 48.3 from 1 benchmark.
Widest category
Reasoning, 27.5 points apart on average over 9 shared benchmarks.
Widest single gap
ARC-AGI-2: Claude Haiku 4.5 4.0% against Gemini 3.6 Flash 60.4%, 56.4 points apart on a 0 to 100 scale.
Output price
Claude Haiku 4.5 $5.00 against Gemini 3.6 Flash $3.75 per million output tokens, 1.3x apart.
Context window
Claude Haiku 4.5 takes 200K tokens, Gemini 3.6 Flash 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 Haiku 4.5BarsGemini 3.6 FlashTies
Coding0121 tie
Intelligence02
Agentic work05
Reasoning09
Math03
Knowledge011
Web design07
General123
All shared boards1721 tie
CodingLike-for-likeFour of these boards feed the Vibe Coding Index012
Coding benchmarks for Claude Haiku 4.5 and Gemini 3.6 Flash
BenchmarkClaude Haiku 4.5Gemini 3.6 FlashLeader
SWE-bench Verified66.6%Not measuredOnly Claude Haiku 4.5 measured
SWE-bench Multilingual64.7%Not measuredOnly Claude Haiku 4.5 measured
LiveBench coding48.1%Not measuredOnly Claude Haiku 4.5 measured
LMArena WebDev13301538Gemini 3.6 Flash
AA Coding Index43.969.2Gemini 3.6 Flash
LiveCodeBench (Vals AI)41.2%88.1%Gemini 3.6 Flash
SWE-bench Verified (Vals AI)66.6%79.6%Gemini 3.6 Flash
Vibe Code Bench (Vals AI)11.4%64.0%Gemini 3.6 Flash
Code migration (Vals AI)7.5%30.9%Gemini 3.6 Flash
ProgramBench (Vals AI)0.0%0.0%Tie
IOI (Vals AI)Not measured35.1%Only Gemini 3.6 Flash measured
Terminal-Bench 2.1 (Vals AI)43.8%73.8%Gemini 3.6 Flash
CyberBench (Vals AI)Not measured45.4%Only Gemini 3.6 Flash measured
SciCodeNot measured53.4%Only Gemini 3.6 Flash measured
Terminal-Bench35.5%Not measuredOnly Claude Haiku 4.5 measured
FrontierCodeNot measured34.4%Only Gemini 3.6 Flash measured
WeirdML45.4%56.1%Gemini 3.6 Flash
ALE-BenchNot measured715.5Only Gemini 3.6 Flash measured
LMArena Text: coding14541491Gemini 3.6 Flash
LMArena Text: software and IT14391492Gemini 3.6 Flash
LMArena WebDev: React13261528Gemini 3.6 Flash
LMArena WebDev: HTML13181547Gemini 3.6 Flash
LMArena WebDev: from an imageNot measured1545Only Gemini 3.6 Flash measured
Neither model has a result on the other 15 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
Intelligence benchmarks for Claude Haiku 4.5 and Gemini 3.6 Flash
BenchmarkClaude Haiku 4.5Gemini 3.6 FlashLeader
AA Intelligence Index16.934.0Gemini 3.6 Flash
Epoch Capabilities Index142.2154.5Gemini 3.6 Flash
Agentic workLike-for-likePublic boards, shown as published, kept out of the index05
Agentic work benchmarks for Claude Haiku 4.5 and Gemini 3.6 Flash
BenchmarkClaude Haiku 4.5Gemini 3.6 FlashLeader
AA Agentic Index8.029.0Gemini 3.6 Flash
Design Arena: agents building web appsNot measured1197Only Gemini 3.6 Flash measured
Design Arena: agents building full-stack appsNot measured1181Only Gemini 3.6 Flash measured
Design Arena: agents building mobile appsNot measured1199Only Gemini 3.6 Flash measured
Design Arena: agents building Android appsNot measured1208Only Gemini 3.6 Flash measured
Design Arena: agents building gamesNot measured1181Only Gemini 3.6 Flash measured
Design Arena: HTML slidesNot measured1148Only Gemini 3.6 Flash measured
Design Arena: PowerPoint slidesNot measured1146Only Gemini 3.6 Flash measured
Design Arena: spreadsheetsNot measured1017Only Gemini 3.6 Flash measured
LiveBench research42.2%Not measuredOnly Claude Haiku 4.5 measured
LiveBench browser40.0%Not measuredOnly Claude Haiku 4.5 measured
LiveBench long agentic11.1%Not measuredOnly Claude Haiku 4.5 measured
Finance Agent (Vals AI)31.0%56.3%Gemini 3.6 Flash
Tax Agent Bench (Vals AI)5.9%18.0%Gemini 3.6 Flash
Legal Research Bench (Vals AI)10.6%25.0%Gemini 3.6 Flash
Excel Modeling Benchmark (Vals AI)22.7%65.4%Gemini 3.6 Flash
BioMysteryBench (Vals AI)Not measured58.5%Only Gemini 3.6 Flash measured
Terminal-Bench-Science (Vals AI)Not measured4.3%Only Gemini 3.6 Flash measured
APEX-AgentsNot measured46.9%Only Gemini 3.6 Flash measured
BALROG31.2%Not measuredOnly Claude Haiku 4.5 measured
DeepResearch Bench45.5%Not measuredOnly Claude Haiku 4.5 measured
Vending-Bench 2$459Not measuredOnly Claude Haiku 4.5 measured
Neither model has a result on the other 17 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index09
Reasoning benchmarks for Claude Haiku 4.5 and Gemini 3.6 Flash
BenchmarkClaude Haiku 4.5Gemini 3.6 FlashLeader
GPQA Diamond72.2%93.4%Gemini 3.6 Flash
LiveBench reasoning61.7%Not measuredOnly Claude Haiku 4.5 measured
MMMU Pro (Vals AI)46.1%88.4%Gemini 3.6 Flash
GPQA Diamond (Artificial Analysis run)Not measured92.8%Only Gemini 3.6 Flash measured
Humanity's Last Exam (Artificial Analysis run)Not measured40.8%Only Gemini 3.6 Flash measured
GPQA Diamond (Epoch AI run)71.2%94.1%Gemini 3.6 Flash
ARC-AGI47.7%91.2%Gemini 3.6 Flash
ARC-AGI-24.0%60.4%Gemini 3.6 Flash
Chess puzzles8.0%42.0%Gemini 3.6 Flash
Mystery game puzzlesNot measured30.0%Only Gemini 3.6 Flash measured
LMArena Text: hard prompts14211486Gemini 3.6 Flash
LMArena Text: expert14431490Gemini 3.6 Flash
LMArena Text: hard prompts in English14351484Gemini 3.6 Flash
LMArena Vision: diagramsNot measured1314Only Gemini 3.6 Flash measured
LMArena Vision: homeworkNot measured1314Only Gemini 3.6 Flash measured
Neither model has a result on the other 6 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index03
Math benchmarks for Claude Haiku 4.5 and Gemini 3.6 Flash
BenchmarkClaude Haiku 4.5Gemini 3.6 FlashLeader
AIME82.7%Not measuredOnly Claude Haiku 4.5 measured
MGSM (Vals AI)92.2%Not measuredOnly Claude Haiku 4.5 measured
OTIS Mock AIME66.7%94.2%Gemini 3.6 Flash
FrontierMath, tiers 1 to 3Not measured58.9%Only Gemini 3.6 Flash measured
FrontierMath, tier 4Not measured22.0%Only Gemini 3.6 Flash measured
ProofBenchNot measured36.0%Only Gemini 3.6 Flash measured
MATH level 596.4%Not measuredOnly Claude Haiku 4.5 measured
LMArena Text: math13981508Gemini 3.6 Flash
LMArena Text: mathematical work14301492Gemini 3.6 Flash
Neither model has a result on the other 4 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index011
Knowledge benchmarks for Claude Haiku 4.5 and Gemini 3.6 Flash
BenchmarkClaude Haiku 4.5Gemini 3.6 FlashLeader
MMLU Pro78.7%89.3%Gemini 3.6 Flash
LegalBench (Vals AI)81.2%86.7%Gemini 3.6 Flash
CaseLaw (Vals AI)56.5%Not measuredOnly Claude Haiku 4.5 measured
CorpFin (Vals AI)60.6%63.3%Gemini 3.6 Flash
TaxEval (Vals AI)67.5%74.9%Gemini 3.6 Flash
MortgageTax (Vals AI)62.2%67.6%Gemini 3.6 Flash
MedQA (Vals AI)79.6%Not measuredOnly Claude Haiku 4.5 measured
MedCode (Vals AI)32.7%53.1%Gemini 3.6 Flash
SimpleQA Verified13.2%66.2%Gemini 3.6 Flash
LMArena Text: science14061489Gemini 3.6 Flash
LMArena Text: medicine and healthcare13941469Gemini 3.6 Flash
LMArena Text: legal and government14001479Gemini 3.6 Flash
LMArena Text: business and finance14001464Gemini 3.6 Flash
Neither model has a result on the other 4 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index07
Web design benchmarks for Claude Haiku 4.5 and Gemini 3.6 Flash
BenchmarkClaude Haiku 4.5Gemini 3.6 FlashLeader
Design Arena: websites11221301Gemini 3.6 Flash
Design Arena: UI components11001297Gemini 3.6 Flash
Design Arena: game dev10971263Gemini 3.6 Flash
Design Arena: data viz11241289Gemini 3.6 Flash
Design Arena: 3D10721272Gemini 3.6 Flash
Design Arena: all code categories11151290Gemini 3.6 Flash
Design Arena: SVG1025Not measuredOnly Claude Haiku 4.5 measured
Design Arena: ASCII art11391281Gemini 3.6 Flash
LiveBench design59.9%Not measuredOnly Claude Haiku 4.5 measured
GeneralLike-for-likePublic boards, shown as published, kept out of the index123
General benchmarks for Claude Haiku 4.5 and Gemini 3.6 Flash
BenchmarkClaude Haiku 4.5Gemini 3.6 FlashLeader
LiveBench data analysis59.3%Not measuredOnly Claude Haiku 4.5 measured
Public Benefits Bench (Vals AI)54.3%56.6%Gemini 3.6 Flash
SAGE (Vals AI)31.8%48.7%Gemini 3.6 Flash
MedScribe (Vals AI)85.2%79.7%Claude Haiku 4.5
Vals Multimodal Index42.9%65.1%Gemini 3.6 Flash
Long context reasoning (AA-LCR)Not measured80.0%Only Gemini 3.6 Flash measured
DTBench73.6%95.5%Gemini 3.6 Flash
LMCA30.9%44.9%Gemini 3.6 Flash
Furniture assemblyNot measured23.3%Only Gemini 3.6 Flash measured
ExploitBench13.7%Not measuredOnly Claude Haiku 4.5 measured
LMArena Text13971480Gemini 3.6 Flash
LMArena Text: instruction following14141467Gemini 3.6 Flash
LMArena Text: creative writing13711466Gemini 3.6 Flash
LMArena VisionNot measured1299Only Gemini 3.6 Flash measured
LMArena Text: English14141482Gemini 3.6 Flash
LMArena Text: non-English13771470Gemini 3.6 Flash
LMArena Text: Chinese14181531Gemini 3.6 Flash
LMArena Text: German13741480Gemini 3.6 Flash
LMArena Text: Japanese13381475Gemini 3.6 Flash
LMArena Text: Korean13461429Gemini 3.6 Flash
LMArena Text: French14111503Gemini 3.6 Flash
LMArena Text: Spanish14211476Gemini 3.6 Flash
LMArena Text: Russian13811488Gemini 3.6 Flash
LMArena Text: Polish13431473Gemini 3.6 Flash
LMArena Text: multi-turn14091481Gemini 3.6 Flash
LMArena Text: longer queries14271475Gemini 3.6 Flash
LMArena Text: writing and language13861470Gemini 3.6 Flash
LMArena Text: entertainment and media13691456Gemini 3.6 Flash
LMArena Vision: EnglishNot measured1295Only Gemini 3.6 Flash measured
LMArena Vision: ChineseNot measured1357Only Gemini 3.6 Flash measured
LMArena Vision: reading textNot measured1312Only Gemini 3.6 Flash measured
LMArena Vision: humorNot measured1289Only Gemini 3.6 Flash measured
LMArena Vision: creative writingNot measured1278Only Gemini 3.6 Flash measured
LMArena Document14201456Gemini 3.6 Flash
Neither model has a result on the other 12 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for Claude Haiku 4.5 and Gemini 3.6 Flash
BenchmarkClaude Haiku 4.5Gemini 3.6 FlashLeader
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 Haiku 4.5$0.0035Fits in one request
Gemini 3.6 Flash$0.0026Fits in one request

Gemini 3.6 Flash costs less

Repository review

50K input + 3K output tokens

Claude Haiku 4.5$0.065Fits in one request
Gemini 3.6 Flash$0.049Fits in one request

Gemini 3.6 Flash costs less

Agent loop

200K input + 10K output tokens

Claude Haiku 4.5$0.250Exceeds the context window
Gemini 3.6 Flash$0.188Fits in one request

Gemini 3.6 Flash 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 Haiku 4.5 and Gemini 3.6 Flash
SpecificationAnthropic Claude Haiku 4.5Google Gemini 3.6 Flash
MakerAnthropicGoogle
WeightsNo public weights listedNo public weights listed
ReleasedOctober 15, 2025July 21, 2026
Context window200K1.05M
Input $/Mtok$1.00$0.75
Output $/Mtok$5.00$3.75
Coding index49.848.3
Independent measurements behind it31
Cost per solved task$0.56$1.49
SourceOpenRouter listingOpenRouter listing

Questions

Is Claude Haiku 4.5 better than Gemini 3.6 Flash for coding?

Claude Haiku 4.5 and Gemini 3.6 Flash are level for coding, 49.8 against 48.3 on the Vibe Coding Index, so price and context decide. Pick Gemini 3.6 Flash when output tokens cost 1.3x less, or when the context window is larger, 1.05M against 200K tokens. Pick Claude Haiku 4.5 when a solved task costs $0.56 against $1.49, or when the benchmark that matters is MedScribe (Vals AI), where it leads.

Which is cheaper, Claude Haiku 4.5 or Gemini 3.6 Flash?

Claude Haiku 4.5 costs $1.00 per million input tokens and $5.00 per million output tokens; Gemini 3.6 Flash costs $0.75 and $3.75. Per solved SWE-bench Verified task, Claude Haiku 4.5 costs $0.56 and Gemini 3.6 Flash $1.49.

Which has the bigger context window, Claude Haiku 4.5 or Gemini 3.6 Flash?

Gemini 3.6 Flash, with 1.05M tokens against 200K.

How many benchmarks is the Claude Haiku 4.5 vs Gemini 3.6 Flash comparison based on?

74 benchmarks have a result for both models as of October 21, 2026. Claude Haiku 4.5's coding index rests on 3 independent benchmarks and Gemini 3.6 Flash's on 1. Where only one model has a result, the page says so instead of naming a winner.

Compare next