Enter opens the highlighted result. Esc closes.

Claude Haiku 4.5 vs Claude Opus 4.8

Claude Opus 4.8 leads Claude Haiku 4.5 on coding, 65.0 to 35.6 on the Vibe Coding Index, and Claude Haiku 4.5 costs 5.0x less per output token ($5.00 vs $25.00 per million). Updated October 22, 2026; 84 benchmarks have a result for both.

Replace Claude Haiku 4.5 Replace Claude Opus 4.8
Claude Haiku 4.5Anthropic
Coding
35.6
Overall
41.0
Output $/M
$5.00
0
Claude Opus 4.8Anthropic
Coding
65.0
Overall
69.8
Output $/M
$25.00
84
vs boards won of 84 shared

Claude Haiku 4.5 or Claude Opus 4.8: which to pick

Pick Claude Opus 4.8 for coding: it scores 65.0 on the Vibe Coding Index against 35.6 for Claude Haiku 4.5.

Pick Claude Haiku 4.5 for output tokens 5.0x cheaper ($5.00 vs $25.00 per million) and a lower cost per solved task ($0.56 against $2.17).

Model AClaude Haiku 4.5Anthropic35.6/1004 independent benchmarks$0.56 per solved taskModel BClaude Opus 4.8Anthropic65.0/1006 independent benchmarks$2.17 per solved task
Shared results
84
Claude Haiku 4.5 only
11
Claude Opus 4.8 only
46
Like-for-like categories
8 of 9

Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.

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 35.6 from 4 benchmarks, Claude Opus 4.8 65.0 from 6 benchmarks.
Widest category
Agentic work, 33.2 points apart on average over 10 shared benchmarks.
Widest single gap
Vibe Code Bench (Vals AI): Claude Haiku 4.5 11.4% against Claude Opus 4.8 82.7%, 71.3 points apart on a 0 to 100 scale.
Output price
Claude Haiku 4.5 $5.00 against Claude Opus 4.8 $25.00 per million output tokens, 5.0x apart.
Context window
Claude Haiku 4.5 takes 200K tokens, Claude Opus 4.8 1M.

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 4.5BarsClaude Opus 4.8Ties
Coding014
Intelligence02
Agentic work010
Reasoning010
Math03
Knowledge011
Web design09
General025
All shared boards084
CodingLike-for-likeSix of these boards feed the Vibe Coding Index014
Coding benchmarks for Claude Haiku 4.5 and Claude Opus 4.8
BenchmarkClaude Haiku 4.5Claude Opus 4.8Leader
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%77.3%Claude Opus 4.8
LMArena WebDev13301556Claude Opus 4.8
AA Coding Index43.974.3Claude Opus 4.8
LiveCodeBench (Vals AI)41.2%87.8%Claude Opus 4.8
SWE-bench Verified (Vals AI)66.6%88.6%Claude Opus 4.8
Terminal-Bench 4 (Vals AI)Not measured23.2%Only Claude Opus 4.8 measured
Vibe Code Bench (Vals AI)11.4%82.7%Claude Opus 4.8
Code migration (Vals AI)7.5%47.3%Claude Opus 4.8
ProgramBench (Vals AI)0.0%1.0%Claude Opus 4.8
Terminal-Bench 2.1 (Vals AI)43.8%71.9%Claude Opus 4.8
SkillsBench (Vals AI)Not measured59.2%Only Claude Opus 4.8 measured
SciCodeNot measured54.4%Only Claude Opus 4.8 measured
Terminal-Bench Hard (Artificial Analysis run)Not measured58.3%Only Claude Opus 4.8 measured
Terminal-Bench35.5%Not measuredOnly Claude Haiku 4.5 measured
FrontierCodeNot measured46.5%Only Claude Opus 4.8 measured
GSO-BenchNot measured47.1%Only Claude Opus 4.8 measured
WeirdML45.4%82.9%Claude Opus 4.8
LMArena Text: coding14541490Claude Opus 4.8
LMArena Text: software and IT14391488Claude Opus 4.8
LMArena WebDev: React13261557Claude Opus 4.8
LMArena WebDev: HTML13181543Claude Opus 4.8
SciCodeNot measured12.3%Only Claude Opus 4.8 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 Claude Opus 4.8
BenchmarkClaude Haiku 4.5Claude Opus 4.8Leader
AA Intelligence Index16.941.8Claude Opus 4.8
Epoch Capabilities Index142.2159.0Claude Opus 4.8
Agentic workLike-for-likePublic boards, shown as published, kept out of the index010
Agentic work benchmarks for Claude Haiku 4.5 and Claude Opus 4.8
BenchmarkClaude Haiku 4.5Claude Opus 4.8Leader
AA Agentic Index8.041.9Claude Opus 4.8
Design Arena: agents building web appsNot measured1233Only Claude Opus 4.8 measured
Design Arena: agents building full-stack appsNot measured1249Only Claude Opus 4.8 measured
Design Arena: agents building mobile appsNot measured1222Only Claude Opus 4.8 measured
Design Arena: agents building Android appsNot measured1233Only Claude Opus 4.8 measured
Design Arena: agents building Godot gamesNot measured1231Only Claude Opus 4.8 measured
Design Arena: agents building gamesNot measured1239Only Claude Opus 4.8 measured
Design Arena: HTML slidesNot measured1171Only Claude Opus 4.8 measured
Design Arena: PowerPoint slidesNot measured1202Only Claude Opus 4.8 measured
Design Arena: spreadsheetsNot measured1303Only Claude Opus 4.8 measured
LiveBench research42.2%81.8%Claude Opus 4.8
LiveBench browser40.0%96.0%Claude Opus 4.8
LiveBench long agentic11.1%48.3%Claude Opus 4.8
Finance Agent (Vals AI)31.0%53.9%Claude Opus 4.8
Tax Agent Bench (Vals AI)5.9%30.0%Claude Opus 4.8
Legal Research Bench (Vals AI)10.6%43.8%Claude Opus 4.8
Excel Modeling Benchmark (Vals AI)22.7%69.4%Claude Opus 4.8
Terminal-Bench-Science (Vals AI)Not measured4.3%Only Claude Opus 4.8 measured
Time Horizon Index (Vals AI)Not measured11.8%Only Claude Opus 4.8 measured
τ²-Bench TelecomNot measured94.4%Only Claude Opus 4.8 measured
APEX-AgentsNot measured48.9%Only Claude Opus 4.8 measured
Remote Labor IndexNot measured8.3%Only Claude Opus 4.8 measured
OSWorld 2.0Not measured20.6%Only Claude Opus 4.8 measured
BALROG31.2%Not measuredOnly Claude Haiku 4.5 measured
DeepResearch Bench45.5%50.2%Claude Opus 4.8
PostTrainBenchNot measured33.8%Only Claude Opus 4.8 measured
Vending-Bench 2$459$5,787Claude Opus 4.8
LMArena SearchNot measured1204Only Claude Opus 4.8 measured
Berkeley Function Calling Leaderboard (BFCL)68.7%Not measuredOnly Claude Haiku 4.5 measured
Neither model has a result on the other 12 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index010
Reasoning benchmarks for Claude Haiku 4.5 and Claude Opus 4.8
BenchmarkClaude Haiku 4.5Claude Opus 4.8Leader
GPQA Diamond72.2%92.4%Claude Opus 4.8
LiveBench reasoning61.7%89.2%Claude Opus 4.8
MMMU Pro (Vals AI)46.1%86.6%Claude Opus 4.8
GPQA Diamond (Artificial Analysis run)Not measured92.0%Only Claude Opus 4.8 measured
Humanity's Last Exam (Artificial Analysis run)Not measured48.7%Only Claude Opus 4.8 measured
GPQA Diamond (Epoch AI run)71.2%91.0%Claude Opus 4.8
ARC-AGI47.7%92.5%Claude Opus 4.8
ARC-AGI-24.0%72.1%Claude Opus 4.8
SimpleBenchNot measured64.8%Only Claude Opus 4.8 measured
Chess puzzles8.0%52.0%Claude Opus 4.8
Mystery game puzzlesNot measured36.0%Only Claude Opus 4.8 measured
EBR-benchNot measured28.6%Only Claude Opus 4.8 measured
LMArena Text: hard prompts14211482Claude Opus 4.8
LMArena Text: expert14431503Claude Opus 4.8
LMArena Text: hard prompts in English14351487Claude Opus 4.8
LMArena Vision: diagramsNot measured1315Only Claude Opus 4.8 measured
LMArena Vision: homeworkNot measured1337Only Claude Opus 4.8 measured
Kagi LLM BenchmarkNot measured88.8%Only Claude Opus 4.8 measured
Neither model has a result on the other 4 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index03
Math benchmarks for Claude Haiku 4.5 and Claude Opus 4.8
BenchmarkClaude Haiku 4.5Claude Opus 4.8Leader
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%98.3%Claude Opus 4.8
FrontierMath, tiers 1 to 3Not measured80.0%Only Claude Opus 4.8 measured
FrontierMath, tier 4Not measured56.1%Only Claude Opus 4.8 measured
ProofBenchNot measured69.0%Only Claude Opus 4.8 measured
MATH level 596.4%Not measuredOnly Claude Haiku 4.5 measured
LMArena Text: math13981486Claude Opus 4.8
LMArena Text: mathematical work14301497Claude Opus 4.8
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 Claude Opus 4.8
BenchmarkClaude Haiku 4.5Claude Opus 4.8Leader
MMLU Pro78.7%89.6%Claude Opus 4.8
LegalBench (Vals AI)81.2%83.6%Claude Opus 4.8
CaseLaw (Vals AI)56.5%Not measuredOnly Claude Haiku 4.5 measured
CorpFin (Vals AI)60.6%66.7%Claude Opus 4.8
TaxEval (Vals AI)67.5%75.6%Claude Opus 4.8
MortgageTax (Vals AI)62.2%69.9%Claude Opus 4.8
MedQA (Vals AI)79.6%Not measuredOnly Claude Haiku 4.5 measured
MedCode (Vals AI)32.7%53.2%Claude Opus 4.8
SimpleQA Verified13.2%53.0%Claude Opus 4.8
LMArena Text: science14061477Claude Opus 4.8
LMArena Text: medicine and healthcare13941464Claude Opus 4.8
LMArena Text: legal and government14001477Claude Opus 4.8
LMArena Text: business and finance14001458Claude Opus 4.8
FACTS Benchmark SuiteNot measured53.5%Only Claude Opus 4.8 measured
Neither model has a result on the other 4 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index09
Web design benchmarks for Claude Haiku 4.5 and Claude Opus 4.8
BenchmarkClaude Haiku 4.5Claude Opus 4.8Leader
Design Arena: websites11221258Claude Opus 4.8
Design Arena: UI components10971255Claude Opus 4.8
Design Arena: game dev10971264Claude Opus 4.8
Design Arena: data viz11241240Claude Opus 4.8
Design Arena: 3D10711222Claude Opus 4.8
Design Arena: all code categories11141255Claude Opus 4.8
Design Arena: SVG10251195Claude Opus 4.8
Design Arena: ASCII art11401277Claude Opus 4.8
LiveBench design59.9%78.8%Claude Opus 4.8
GeneralLike-for-likePublic boards, shown as published, kept out of the index025
General benchmarks for Claude Haiku 4.5 and Claude Opus 4.8
BenchmarkClaude Haiku 4.5Claude Opus 4.8Leader
LiveBench data analysis59.3%66.0%Claude Opus 4.8
Public Benefits Bench (Vals AI)54.3%68.1%Claude Opus 4.8
SAGE (Vals AI)31.8%54.8%Claude Opus 4.8
MedScribe (Vals AI)85.2%85.8%Claude Opus 4.8
Vals Multimodal Index42.9%70.9%Claude Opus 4.8
IFBenchNot measured62.2%Only Claude Opus 4.8 measured
Long context reasoning (AA-LCR)Not measured77.7%Only Claude Opus 4.8 measured
DTBench73.6%94.9%Claude Opus 4.8
LMCA30.9%57.5%Claude Opus 4.8
Surface Evolver BenchNot measured87.5%Only Claude Opus 4.8 measured
Furniture assemblyNot measured42.5%Only Claude Opus 4.8 measured
ExploitBench13.7%Not measuredOnly Claude Haiku 4.5 measured
LMArena Text13971461Claude Opus 4.8
LMArena Text: instruction following14141476Claude Opus 4.8
LMArena Text: creative writing13711454Claude Opus 4.8
LMArena VisionNot measured1294Only Claude Opus 4.8 measured
LMArena Text: English14141468Claude Opus 4.8
LMArena Text: non-English13771450Claude Opus 4.8
LMArena Text: Chinese14181506Claude Opus 4.8
LMArena Text: German13741472Claude Opus 4.8
LMArena Text: Japanese13381442Claude Opus 4.8
LMArena Text: Korean13461431Claude Opus 4.8
LMArena Text: French14111481Claude Opus 4.8
LMArena Text: Spanish14211465Claude Opus 4.8
LMArena Text: Russian13811474Claude Opus 4.8
LMArena Text: Polish13431459Claude Opus 4.8
LMArena Text: multi-turn14091476Claude Opus 4.8
LMArena Text: longer queries14271483Claude Opus 4.8
LMArena Text: writing and language13861461Claude Opus 4.8
LMArena Text: entertainment and media13691440Claude Opus 4.8
LMArena Vision: EnglishNot measured1293Only Claude Opus 4.8 measured
LMArena Vision: ChineseNot measured1346Only Claude Opus 4.8 measured
LMArena Vision: reading textNot measured1303Only Claude Opus 4.8 measured
LMArena Vision: humorNot measured1282Only Claude Opus 4.8 measured
LMArena Vision: creative writingNot measured1301Only Claude Opus 4.8 measured
LMArena Document14201475Claude Opus 4.8
Neither model has a result on the other 10 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 Claude Opus 4.8
BenchmarkClaude Haiku 4.5Claude Opus 4.8Leader
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
Claude Opus 4.8$0.018Fits 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
Claude Opus 4.8$0.325Fits 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
Claude Opus 4.8$1.25Fits 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 Claude Opus 4.8
SpecificationAnthropic Claude Haiku 4.5Anthropic Claude Opus 4.8
MakerAnthropicAnthropic
WeightsNo public weights listedNo public weights listed
ReleasedOctober 15, 2025May 27, 2026
Context window200K1M
Input $/Mtok$1.00$5.00
Output $/Mtok$5.00$25.00
Coding index35.665.0
Independent measurements behind it46
Cost per solved task$0.56$2.17
SourceOpenRouter listingOpenRouter listing

Questions

Is Claude Haiku 4.5 better than Claude Opus 4.8 for coding?

Pick Claude Opus 4.8 for coding: it scores 65.0 on the Vibe Coding Index against 35.6 for Claude Haiku 4.5. Pick Claude Haiku 4.5 for output tokens 5.0x cheaper ($5.00 vs $25.00 per million) and a lower cost per solved task ($0.56 against $2.17).

Which is cheaper, Claude Haiku 4.5 or Claude Opus 4.8?

Claude Haiku 4.5 costs $1.00 per million input tokens and $5.00 per million output tokens; Claude Opus 4.8 costs $5.00 and $25.00. Per solved SWE-bench Verified task, Claude Haiku 4.5 costs $0.56 and Claude Opus 4.8 $2.17.

Which has the bigger context window, Claude Haiku 4.5 or Claude Opus 4.8?

Claude Opus 4.8, with 1M tokens against 200K.

How many benchmarks is the Claude Haiku 4.5 vs Claude Opus 4.8 comparison based on?

84 benchmarks have a result for both models as of October 22, 2026. Claude Haiku 4.5's coding index rests on 4 independent benchmarks and Claude Opus 4.8's on 6. 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-claude-opus-4-8" title="Claude Haiku 4.5 vs Claude Opus 4.8, 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-claude-opus-4-8">LLMBenchmark.ai, Claude Haiku 4.5 vs Claude Opus 4.8</a></p>