LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

Claude Haiku 4.5 vs GPT-5.2

Updated October 21, 2026. 72 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 GPT-5.2
Claude Haiku 4.5Anthropic
Coding
49.8
Overall
16.9
Output $/M
$5.00
9
GPT-5.2OpenAI
Coding
58.9
Overall
30.4
Output $/M
$14.00
63
vs boards won of 72 shared

Decision reading

Pick GPT-5.2 for coding: it scores 58.9 on the Vibe Coding Index against 49.8 for Claude Haiku 4.5, though on 2 benchmarks against 3.

Pick Claude Haiku 4.5 when output tokens cost 2.8x less, or when a solved task costs $0.56 against $1.91.

Model AClaude Haiku 4.5Anthropic49.8/1003 independent benchmarks$0.56 per solved taskModel BGPT-5.2OpenAI58.9/1002 independent benchmarks$1.91 per solved task
Shared results
72
Claude Haiku 4.5 only
22
GPT-5.2 only
39
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, GPT-5.2 58.9 from 2 benchmarks.
Widest category
Reasoning, 23.3 points apart on average over 9 shared benchmarks.
Widest single gap
ARC-AGI-2: Claude Haiku 4.5 4.0% against GPT-5.2 52.9%, 48.9 points apart on a 0 to 100 scale.
Output price
Claude Haiku 4.5 $5.00 against GPT-5.2 $14.00 per million output tokens, 2.8x apart.
Context window
Claude Haiku 4.5 takes 200K tokens, GPT-5.2 400K.

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 4.5BarsGPT-5.2Ties
Coding29
Intelligence02
Agentic work11
Reasoning18
Math05
Knowledge013
Web design08
General517
All shared boards963
CodingLike-for-likeFour of these boards feed the Vibe Coding Index29
Coding benchmarks for Claude Haiku 4.5 and GPT-5.2
BenchmarkClaude Haiku 4.5GPT-5.2Leader
SWE-bench Verified66.6%72.8%GPT-5.2
SWE-bench Multilingual64.7%66.7%GPT-5.2
LiveBench coding48.1%Not measuredOnly Claude Haiku 4.5 measured
LMArena WebDev13301417GPT-5.2
AA Coding Index43.9Not measuredOnly Claude Haiku 4.5 measured
LiveCodeBench (Vals AI)41.2%85.4%GPT-5.2
SWE-bench Verified (Vals AI)66.6%75.8%GPT-5.2
Vibe Code Bench (Vals AI)11.4%53.5%GPT-5.2
Code migration (Vals AI)7.5%Not measuredOnly Claude Haiku 4.5 measured
ProgramBench (Vals AI)0.0%Not measuredOnly Claude Haiku 4.5 measured
Terminal-Bench 2.1 (Vals AI)43.8%Not measuredOnly Claude Haiku 4.5 measured
LiveCodeBench (Artificial Analysis run)Not measured89.4%Only GPT-5.2 measured
Terminal-Bench Hard (Artificial Analysis run)Not measured47.0%Only GPT-5.2 measured
Terminal-Bench35.5%64.9%GPT-5.2
GSO-BenchNot measured27.5%Only GPT-5.2 measured
WeirdML45.4%72.2%GPT-5.2
ALE-BenchNot measured1293.5Only GPT-5.2 measured
LMArena Text: coding14541442Claude Haiku 4.5
LMArena Text: software and IT14391434Claude Haiku 4.5
LMArena WebDev: React1326Not measuredOnly Claude Haiku 4.5 measured
LMArena WebDev: HTML13181427GPT-5.2
Neither model has a result on the other 17 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
Intelligence benchmarks for Claude Haiku 4.5 and GPT-5.2
BenchmarkClaude Haiku 4.5GPT-5.2Leader
AA Intelligence Index16.930.4GPT-5.2
Epoch Capabilities Index142.2153.7GPT-5.2
Agentic workLike-for-likePublic boards, shown as published, kept out of the index11
Agentic work benchmarks for Claude Haiku 4.5 and GPT-5.2
BenchmarkClaude Haiku 4.5GPT-5.2Leader
AA Agentic Index8.0Not measuredOnly Claude Haiku 4.5 measured
Design Arena: agents building web appsNot measured1086Only GPT-5.2 measured
Design Arena: agents building full-stack appsNot measured1043Only GPT-5.2 measured
Design Arena: agents building mobile appsNot measured1094Only GPT-5.2 measured
Design Arena: agents building Android appsNot measured1064Only GPT-5.2 measured
Design Arena: agents building Godot gamesNot measured1142Only GPT-5.2 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%Not measuredOnly Claude Haiku 4.5 measured
Tax Agent Bench (Vals AI)5.9%Not measuredOnly Claude Haiku 4.5 measured
Legal Research Bench (Vals AI)10.6%Not measuredOnly Claude Haiku 4.5 measured
Excel Modeling Benchmark (Vals AI)22.7%Not measuredOnly Claude Haiku 4.5 measured
τ²-Bench TelecomNot measured84.8%Only GPT-5.2 measured
METR time horizonsNot measured75.3%Only GPT-5.2 measured
Remote Labor IndexNot measured2.5%Only GPT-5.2 measured
GDPvalNot measured49.7%Only GPT-5.2 measured
BALROG31.2%Not measuredOnly Claude Haiku 4.5 measured
DeepResearch Bench45.5%41.1%Claude Haiku 4.5
Vending-Bench 2$459$3,591GPT-5.2
Neither model has a result on the other 19 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index18
Reasoning benchmarks for Claude Haiku 4.5 and GPT-5.2
BenchmarkClaude Haiku 4.5GPT-5.2Leader
GPQA Diamond72.2%91.7%GPT-5.2
LiveBench reasoning61.7%Not measuredOnly Claude Haiku 4.5 measured
MMMU Pro (Vals AI)46.1%86.7%GPT-5.2
GPQA Diamond (Artificial Analysis run)Not measured90.3%Only GPT-5.2 measured
Humanity's Last Exam (Artificial Analysis run)Not measured37.7%Only GPT-5.2 measured
GPQA Diamond (Epoch AI run)71.2%91.4%GPT-5.2
ARC-AGI47.7%86.2%GPT-5.2
ARC-AGI-24.0%52.9%GPT-5.2
SimpleBenchNot measured45.8%Only GPT-5.2 measured
Chess puzzles8.0%48.0%GPT-5.2
Mystery game puzzlesNot measured23.0%Only GPT-5.2 measured
EBR-benchNot measured23.0%Only GPT-5.2 measured
VPCTNot measured84.0%Only GPT-5.2 measured
LMArena Text: hard prompts14211428GPT-5.2
LMArena Text: expert14431445GPT-5.2
LMArena Text: hard prompts in English14351431Claude Haiku 4.5
LMArena Vision: diagramsNot measured1265Only GPT-5.2 measured
LMArena Vision: homeworkNot measured1276Only GPT-5.2 measured
Neither model has a result on the other 3 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index05
Math benchmarks for Claude Haiku 4.5 and GPT-5.2
BenchmarkClaude Haiku 4.5GPT-5.2Leader
AIME82.7%96.9%GPT-5.2
MGSM (Vals AI)92.2%94.0%GPT-5.2
AA Math IndexNot measured99.0Only GPT-5.2 measured
AIME 2025 (Artificial Analysis run)Not measured99.0%Only GPT-5.2 measured
OTIS Mock AIME66.7%96.1%GPT-5.2
FrontierMath, tiers 1 to 3Not measured67.4%Only GPT-5.2 measured
FrontierMath, tier 4Not measured31.7%Only GPT-5.2 measured
ProofBenchNot measured15.0%Only GPT-5.2 measured
MATH level 596.4%Not measuredOnly Claude Haiku 4.5 measured
LMArena Text: math13981440GPT-5.2
LMArena Text: mathematical work14301442GPT-5.2
Neither model has a result on the other 2 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index013
Knowledge benchmarks for Claude Haiku 4.5 and GPT-5.2
BenchmarkClaude Haiku 4.5GPT-5.2Leader
MMLU Pro78.7%86.2%GPT-5.2
LegalBench (Vals AI)81.2%82.8%GPT-5.2
CaseLaw (Vals AI)56.5%66.0%GPT-5.2
CorpFin (Vals AI)60.6%65.9%GPT-5.2
TaxEval (Vals AI)67.5%75.8%GPT-5.2
MortgageTax (Vals AI)62.2%67.1%GPT-5.2
MedQA (Vals AI)79.6%94.1%GPT-5.2
MedCode (Vals AI)32.7%49.8%GPT-5.2
MMLU Pro (Artificial Analysis run)Not measured87.4%Only GPT-5.2 measured
SimpleQA Verified13.2%37.1%GPT-5.2
LMArena Text: science14061425GPT-5.2
LMArena Text: medicine and healthcare13941425GPT-5.2
LMArena Text: legal and government14001427GPT-5.2
LMArena Text: business and finance14001414GPT-5.2
Neither model has a result on the other 3 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index08
Web design benchmarks for Claude Haiku 4.5 and GPT-5.2
BenchmarkClaude Haiku 4.5GPT-5.2Leader
Design Arena: websites11221196GPT-5.2
Design Arena: UI components11001190GPT-5.2
Design Arena: game dev10971196GPT-5.2
Design Arena: data viz11241202GPT-5.2
Design Arena: 3D10721079GPT-5.2
Design Arena: all code categories11151171GPT-5.2
Design Arena: SVG10251137GPT-5.2
Design Arena: ASCII art11391153GPT-5.2
LiveBench design59.9%Not measuredOnly Claude Haiku 4.5 measured
GeneralLike-for-likePublic boards, shown as published, kept out of the index517
General benchmarks for Claude Haiku 4.5 and GPT-5.2
BenchmarkClaude Haiku 4.5GPT-5.2Leader
LiveBench data analysis59.3%Not measuredOnly Claude Haiku 4.5 measured
Public Benefits Bench (Vals AI)54.3%Not measuredOnly Claude Haiku 4.5 measured
SAGE (Vals AI)31.8%49.3%GPT-5.2
MedScribe (Vals AI)85.2%84.4%Claude Haiku 4.5
Vals Multimodal Index42.9%Not measuredOnly Claude Haiku 4.5 measured
IFBenchNot measured75.4%Only GPT-5.2 measured
Long context reasoning (AA-LCR)Not measured82.7%Only GPT-5.2 measured
DTBench73.6%90.9%GPT-5.2
LMCA30.9%43.9%GPT-5.2
CL-benchNot measured18.2%Only GPT-5.2 measured
Furniture assemblyNot measured38.3%Only GPT-5.2 measured
ExploitBench13.7%Not measuredOnly Claude Haiku 4.5 measured
LMArena Text13971416GPT-5.2
LMArena Text: instruction following14141409Claude Haiku 4.5
LMArena Text: creative writing13711376GPT-5.2
LMArena VisionNot measured1246Only GPT-5.2 measured
LMArena Text: English14141422GPT-5.2
LMArena Text: non-English13771405GPT-5.2
LMArena Text: Chinese14181447GPT-5.2
LMArena Text: German13741444GPT-5.2
LMArena Text: Japanese13381420GPT-5.2
LMArena Text: Korean13461374GPT-5.2
LMArena Text: French14111454GPT-5.2
LMArena Text: Spanish14211413Claude Haiku 4.5
LMArena Text: Russian13811415GPT-5.2
LMArena Text: Polish13431426GPT-5.2
LMArena Text: multi-turn14091419GPT-5.2
LMArena Text: longer queries14271413Claude Haiku 4.5
LMArena Text: writing and language13861392GPT-5.2
LMArena Text: entertainment and media13691378GPT-5.2
LMArena Vision: EnglishNot measured1243Only GPT-5.2 measured
LMArena Vision: ChineseNot measured1305Only GPT-5.2 measured
LMArena Vision: reading textNot measured1257Only GPT-5.2 measured
LMArena Vision: humorNot measured1227Only GPT-5.2 measured
LMArena Vision: captioningNot measured1267Only GPT-5.2 measured
LMArena Vision: recognizing thingsNot measured1232Only GPT-5.2 measured
LMArena Vision: creative writingNot measured1225Only GPT-5.2 measured
LMArena Document14201405Claude Haiku 4.5
Neither model has a result on the other 8 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 GPT-5.2
BenchmarkClaude Haiku 4.5GPT-5.2Leader
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
GPT-5.2$0.0088Fits in one request

Claude Haiku 4.5 costs less

Repository review

50K input + 3K output tokens

Claude Haiku 4.5$0.065Fits in one request
GPT-5.2$0.130Fits in one request

Claude Haiku 4.5 costs less

Agent loop

200K input + 10K output tokens

Claude Haiku 4.5$0.250Exceeds the context window
GPT-5.2$0.490Fits in one request

Claude Haiku 4.5 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 GPT-5.2
SpecificationAnthropic Claude Haiku 4.5OpenAI GPT-5.2
MakerAnthropicOpenAI
WeightsNo public weights listedNo public weights listed
ReleasedOctober 15, 2025December 10, 2025
Context window200K400K
Input $/Mtok$1.00$1.75
Output $/Mtok$5.00$14.00
Coding index49.858.9
Independent measurements behind it32
Cost per solved task$0.56$1.91
SourceOpenRouter listingOpenRouter listing

Questions

Is Claude Haiku 4.5 better than GPT-5.2 for coding?

Pick GPT-5.2 for coding: it scores 58.9 on the Vibe Coding Index against 49.8 for Claude Haiku 4.5, though on 2 benchmarks against 3. Pick Claude Haiku 4.5 when output tokens cost 2.8x less, or when a solved task costs $0.56 against $1.91.

Which is cheaper, Claude Haiku 4.5 or GPT-5.2?

Claude Haiku 4.5 costs $1.00 per million input tokens and $5.00 per million output tokens; GPT-5.2 costs $1.75 and $14.00. Per solved SWE-bench Verified task, Claude Haiku 4.5 costs $0.56 and GPT-5.2 $1.91.

Which has the bigger context window, Claude Haiku 4.5 or GPT-5.2?

GPT-5.2, with 400K tokens against 200K.

How many benchmarks is the Claude Haiku 4.5 vs GPT-5.2 comparison based on?

72 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 GPT-5.2's on 2. Where only one model has a result, the page says so instead of naming a winner.

Compare next

Embed this comparison on your site

Free under CC BY 4.0, while the source link stays visible. Preview the frame · Terms and other embeds

<iframe src="https://llmbenchmark.ai/embed/compare/claude-haiku-4-5-vs-gpt-5-2" title="Claude Haiku 4.5 vs GPT-5.2, LLMBenchmark.ai" width="100%" height="360" style="border:0;max-width:100%" loading="lazy"></iframe>
<p style="font:13px/1.4 sans-serif;margin:6px 0 0">Source: <a href="https://llmbenchmark.ai/compare/claude-haiku-4-5-vs-gpt-5-2">LLMBenchmark.ai, Claude Haiku 4.5 vs GPT-5.2</a></p>