Claude Haiku 5.5 vs Claude Opus 4.6
Claude Haiku 5.5 leads Claude Opus 4.6 on coding, 70.7 to 57.2 on the Vibe Coding Index, and Claude Haiku 5.5 costs 50.0x less per output token ($0.50 vs $25.00 per million). Updated October 22, 2026; 34 benchmarks have a result for both.
vs boards won of 34 shared
Claude Haiku 5.5 or Claude Opus 4.6: which to pick
Pick Claude Haiku 5.5 for coding: it scores 70.7 on the Vibe Coding Index against 57.2 for Claude Opus 4.6.
Pick Claude Opus 4.6 for work measured by Vending-Bench 2 or GPQA Diamond (Epoch AI run), where it leads.
Model AClaude Haiku 5.5Anthropic70.7/1005 independent benchmarksNo cost per solved taskModel BClaude Opus 4.6Anthropic57.2/1004 independent benchmarks$1.56 per solved task
- Shared results
- 34
- Claude Haiku 5.5 only
- 29
- Claude Opus 4.6 only
- 76
- 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.
Coding
Vibe Coding Index, six public coding benchmarks
Claude Haiku 5.5
Claude Haiku 5.5 scores 70.7 against 57.2.
5 and 4 benchmarks behind the scores
Intelligence
AA Intelligence Index
Claude Haiku 5.5
Claude Haiku 5.5 leads, 43.4 against 31.9.
Same board, both measured
Reasoning
GPQA Diamond
No call
Only Claude Opus 4.6 has a GPQA Diamond result, 89.7%.
Directional only
Math
AIME
No call
Only Claude Opus 4.6 has a AIME result, 95.6%.
Directional only
Knowledge
MMLU Pro
No call
Only Claude Opus 4.6 has a MMLU Pro result, 89.1%.
Directional only
Web design
Design Arena: websites
Claude Opus 4.6
Claude Opus 4.6 leads, 1291 against 1262.
Same board, both measured
General
Vals Index
No call
Only Claude Haiku 5.5 has a Vals Index result, 58.7%.
Directional only
Long documents
Listed context window
No call
Both take 1M tokens.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Claude Haiku 5.5
Claude Haiku 5.5 $0.0003, Claude Opus 4.6 $0.018 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only Claude Opus 4.6 has the number, $1.56.
Directional only
At a glance
- Coding index
- Claude Haiku 5.5 70.7 from 5 benchmarks, Claude Opus 4.6 57.2 from 4 benchmarks.
- Widest category
- Agentic work, 18.7 points apart on average over 2 shared benchmarks.
- Widest single gap
- ProofBench: Claude Haiku 5.5 89.0% against Claude Opus 4.6 50.0%, 39.0 points apart on a 0 to 100 scale.
- Output price
- Claude Haiku 5.5 $0.50 against Claude Opus 4.6 $25.00 per million output tokens, 50.0x apart.
- Context window
- Both take 1M tokens.
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.
| Category | Claude Haiku 5.5 | Bars | Claude Opus 4.6 | Ties |
|---|---|---|---|---|
| Coding | 5 | 0 | ||
| Intelligence | 2 | 0 | ||
| Agentic work | 1 | 1 | ||
| Reasoning | 4 | 3 | ||
| Math | 4 | 0 | ||
| Knowledge | 1 | 1 | ||
| Web design | 4 | 2 | ||
| General | 4 | 2 | ||
| All shared boards | 25 | 9 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index50
| Benchmark | Claude Haiku 5.5 | Claude Opus 4.6 | Leader |
|---|---|---|---|
| SWE-bench Verified | Not measured | 75.6% | Only Claude Opus 4.6 measured |
| SWE-bench Multilingual | Not measured | 72.0% | Only Claude Opus 4.6 measured |
| LiveBench coding | 91.4% | Not measured | Only Claude Haiku 5.5 measured |
| LMArena WebDev | 1586 | 1547 | Claude Haiku 5.5 |
| LiveCodeBench (Vals AI) | Not measured | 84.7% | Only Claude Opus 4.6 measured |
| SWE-bench Verified (Vals AI) | Not measured | 78.2% | Only Claude Opus 4.6 measured |
| Terminal-Bench 4 (Vals AI) | 36.4% | Not measured | Only Claude Haiku 5.5 measured |
| Vibe Code Bench (Vals AI) | 90.4% | 57.6% | Claude Haiku 5.5 |
| Code migration (Vals AI) | 44.6% | Not measured | Only Claude Haiku 5.5 measured |
| ProgramBench (Vals AI) | 2.5% | Not measured | Only Claude Haiku 5.5 measured |
| IOI (Vals AI) | 47.3% | Not measured | Only Claude Haiku 5.5 measured |
| SWE-bench Multimodal (Vals AI) | 44.0% | Not measured | Only Claude Haiku 5.5 measured |
| Vibe Code Bench 1-100 (Vals AI) | 20.4% | Not measured | Only Claude Haiku 5.5 measured |
| CyberBench (Vals AI) | 75.5% | Not measured | Only Claude Haiku 5.5 measured |
| SciCode | 55.0% | Not measured | Only Claude Haiku 5.5 measured |
| Terminal-Bench Hard (Artificial Analysis run) | Not measured | 48.5% | Only Claude Opus 4.6 measured |
| Terminal-Bench | Not measured | 79.8% | Only Claude Opus 4.6 measured |
| FrontierCode | 46.4% | 26.6% | Claude Haiku 5.5 |
| FrontierSWE | 43.8% | Not measured | Only Claude Haiku 5.5 measured |
| GSO-Bench | Not measured | 41.2% | Only Claude Opus 4.6 measured |
| WeirdML | Not measured | 78.0% | Only Claude Opus 4.6 measured |
| Cybench | Not measured | 93.0% | Only Claude Opus 4.6 measured |
| LMArena Text: coding | Not measured | 1534 | Only Claude Opus 4.6 measured |
| LMArena Text: software and IT | Not measured | 1527 | Only Claude Opus 4.6 measured |
| LMArena WebDev: React | 1605 | 1539 | Claude Haiku 5.5 |
| LMArena WebDev: HTML | 1569 | 1559 | Claude Haiku 5.5 |
| LMArena WebDev: from an image | Not measured | 1543 | Only Claude Opus 4.6 measured |
| SciCode | 9.2% | Not measured | Only Claude Haiku 5.5 measured |
| Neither model has a result on the other 11 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | Claude Haiku 5.5 | Claude Opus 4.6 | Leader |
|---|---|---|---|
| AA Intelligence Index | 43.4 | 31.9 | Claude Haiku 5.5 |
| Epoch Capabilities Index | 159.3 | 155.7 | Claude Haiku 5.5 |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index11
| Benchmark | Claude Haiku 5.5 | Claude Opus 4.6 | Leader |
|---|---|---|---|
| Design Arena: agents building web apps | Not measured | 1206 | Only Claude Opus 4.6 measured |
| Design Arena: agents building full-stack apps | Not measured | 1220 | Only Claude Opus 4.6 measured |
| Design Arena: agents building mobile apps | Not measured | 1207 | Only Claude Opus 4.6 measured |
| Design Arena: agents building Android apps | Not measured | 1197 | Only Claude Opus 4.6 measured |
| Design Arena: agents building games | Not measured | 1212 | Only Claude Opus 4.6 measured |
| LiveBench research | 81.9% | Not measured | Only Claude Haiku 5.5 measured |
| LiveBench browser | 80.0% | Not measured | Only Claude Haiku 5.5 measured |
| LiveBench long agentic | 57.6% | Not measured | Only Claude Haiku 5.5 measured |
| SREBench (Vals AI) | 5.7% | Not measured | Only Claude Haiku 5.5 measured |
| Finance Agent (Vals AI) | 54.1% | Not measured | Only Claude Haiku 5.5 measured |
| Tax Agent Bench (Vals AI) | 31.9% | Not measured | Only Claude Haiku 5.5 measured |
| Legal Research Bench (Vals AI) | 43.3% | Not measured | Only Claude Haiku 5.5 measured |
| Harvey's Legal Agent Benchmark (Vals AI) | 4.2% | Not measured | Only Claude Haiku 5.5 measured |
| Excel Modeling Benchmark (Vals AI) | 59.0% | Not measured | Only Claude Haiku 5.5 measured |
| BioMysteryBench (Vals AI) | 70.4% | Not measured | Only Claude Haiku 5.5 measured |
| Terminal-Bench-Science (Vals AI) | 10.0% | Not measured | Only Claude Haiku 5.5 measured |
| Vals Science Index | 38.5% | Not measured | Only Claude Haiku 5.5 measured |
| τ²-Bench Telecom | Not measured | 92.1% | Only Claude Opus 4.6 measured |
| APEX-Agents | 56.1% | 46.3% | Claude Haiku 5.5 |
| METR time horizons | Not measured | 78.9% | Only Claude Opus 4.6 measured |
| Remote Labor Index | Not measured | 4.2% | Only Claude Opus 4.6 measured |
| DeepResearch Bench | Not measured | 55.3% | Only Claude Opus 4.6 measured |
| Vending-Bench 2 | $3,602 | $8,018 | Claude Opus 4.6 |
| Neither model has a result on the other 18 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index43
| Benchmark | Claude Haiku 5.5 | Claude Opus 4.6 | Leader |
|---|---|---|---|
| GPQA Diamond | Not measured | 89.7% | Only Claude Opus 4.6 measured |
| LiveBench reasoning | 82.4% | Not measured | Only Claude Haiku 5.5 measured |
| MysteryMechanism (Vals AI) | 35.1% | Not measured | Only Claude Haiku 5.5 measured |
| MMMU Pro (Vals AI) | Not measured | 83.9% | Only Claude Opus 4.6 measured |
| GPQA Diamond (Artificial Analysis run) | Not measured | 89.6% | Only Claude Opus 4.6 measured |
| Humanity's Last Exam (Artificial Analysis run) | 44.4% | 39.9% | Claude Haiku 5.5 |
| GPQA Diamond (Epoch AI run) | 89.6% | 90.5% | Claude Opus 4.6 |
| ARC-AGI | 89.0% | 94.0% | Claude Opus 4.6 |
| ARC-AGI-2 | 61.3% | 69.2% | Claude Opus 4.6 |
| SimpleBench | Not measured | 67.6% | Only Claude Opus 4.6 measured |
| Chess puzzles | 55.0% | 21.0% | Claude Haiku 5.5 |
| Mystery game puzzles | 30.0% | 25.0% | Claude Haiku 5.5 |
| EBR-bench | 43.8% | 12.7% | Claude Haiku 5.5 |
| LMArena Text: hard prompts | Not measured | 1525 | Only Claude Opus 4.6 measured |
| LMArena Text: expert | Not measured | 1544 | Only Claude Opus 4.6 measured |
| LMArena Text: hard prompts in English | Not measured | 1530 | Only Claude Opus 4.6 measured |
| LMArena Vision: diagrams | Not measured | 1340 | Only Claude Opus 4.6 measured |
| LMArena Vision: homework | Not measured | 1337 | Only Claude Opus 4.6 measured |
| Kagi LLM Benchmark | Not measured | 83.6% | Only Claude Opus 4.6 measured |
| Neither model has a result on the other 3 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index40
| Benchmark | Claude Haiku 5.5 | Claude Opus 4.6 | Leader |
|---|---|---|---|
| AIME | Not measured | 95.6% | Only Claude Opus 4.6 measured |
| OTIS Mock AIME | 98.9% | 94.4% | Claude Haiku 5.5 |
| FrontierMath, tiers 1 to 3 | 75.1% | 66.0% | Claude Haiku 5.5 |
| FrontierMath, tier 4 | 46.3% | 26.8% | Claude Haiku 5.5 |
| ProofBench | 89.0% | 50.0% | Claude Haiku 5.5 |
| LMArena Text: math | Not measured | 1518 | Only Claude Opus 4.6 measured |
| LMArena Text: mathematical work | Not measured | 1525 | Only Claude Opus 4.6 measured |
| Neither model has a result on the other 6 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index11
| Benchmark | Claude Haiku 5.5 | Claude Opus 4.6 | Leader |
|---|---|---|---|
| MMLU Pro | Not measured | 89.1% | Only Claude Opus 4.6 measured |
| LegalBench (Vals AI) | Not measured | 85.3% | Only Claude Opus 4.6 measured |
| CaseLaw (Vals AI) | Not measured | 62.1% | Only Claude Opus 4.6 measured |
| CorpFin (Vals AI) | Not measured | 67.0% | Only Claude Opus 4.6 measured |
| TaxEval (Vals AI) | Not measured | 76.0% | Only Claude Opus 4.6 measured |
| MortgageTax (Vals AI) | Not measured | 68.5% | Only Claude Opus 4.6 measured |
| MedQA (Vals AI) | Not measured | 95.4% | Only Claude Opus 4.6 measured |
| MedCode (Vals AI) | 51.5% | 49.1% | Claude Haiku 5.5 |
| SimpleQA Verified | 23.8% | 47.0% | Claude Opus 4.6 |
| LMArena Text: science | Not measured | 1519 | Only Claude Opus 4.6 measured |
| LMArena Text: medicine and healthcare | Not measured | 1504 | Only Claude Opus 4.6 measured |
| LMArena Text: legal and government | Not measured | 1512 | Only Claude Opus 4.6 measured |
| LMArena Text: business and finance | Not measured | 1497 | Only Claude Opus 4.6 measured |
| FACTS Benchmark Suite | Not measured | 41.5% | Only Claude Opus 4.6 measured |
| Neither model has a result on the other 4 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index42
| Benchmark | Claude Haiku 5.5 | Claude Opus 4.6 | Leader |
|---|---|---|---|
| Design Arena: websites | 1262 | 1291 | Claude Opus 4.6 |
| Design Arena: UI components | 1291 | 1281 | Claude Haiku 5.5 |
| Design Arena: game dev | 1313 | 1277 | Claude Haiku 5.5 |
| Design Arena: data viz | 1294 | 1279 | Claude Haiku 5.5 |
| Design Arena: 3D | 1342 | 1273 | Claude Haiku 5.5 |
| Design Arena: all code categories | 1278 | 1288 | Claude Opus 4.6 |
| Design Arena: SVG | Not measured | 1227 | Only Claude Opus 4.6 measured |
| Design Arena: ASCII art | Not measured | 1255 | Only Claude Opus 4.6 measured |
| LiveBench design | 83.1% | Not measured | Only Claude Haiku 5.5 measured |
GeneralLike-for-likePublic boards, shown as published, kept out of the index42
| Benchmark | Claude Haiku 5.5 | Claude Opus 4.6 | Leader |
|---|---|---|---|
| Vals Index | 58.7% | Not measured | Only Claude Haiku 5.5 measured |
| LiveBench data analysis | 53.1% | Not measured | Only Claude Haiku 5.5 measured |
| Public Benefits Bench (Vals AI) | 63.2% | Not measured | Only Claude Haiku 5.5 measured |
| SAGE (Vals AI) | 45.8% | 51.6% | Claude Opus 4.6 |
| MedScribe (Vals AI) | 89.5% | 86.7% | Claude Haiku 5.5 |
| IFBench | Not measured | 53.1% | Only Claude Opus 4.6 measured |
| Long context reasoning (AA-LCR) | 82.7% | 78.0% | Claude Haiku 5.5 |
| DTBench | 95.0% | 91.2% | Claude Haiku 5.5 |
| LMCA | 51.1% | 55.8% | Claude Opus 4.6 |
| CL-bench | Not measured | 20.7% | Only Claude Opus 4.6 measured |
| Furniture assembly | 47.5% | 28.3% | Claude Haiku 5.5 |
| CL-bench Life | Not measured | 17.0% | Only Claude Opus 4.6 measured |
| LMArena Text | Not measured | 1501 | Only Claude Opus 4.6 measured |
| LMArena Text: instruction following | Not measured | 1521 | Only Claude Opus 4.6 measured |
| LMArena Text: creative writing | Not measured | 1505 | Only Claude Opus 4.6 measured |
| LMArena Vision | Not measured | 1316 | Only Claude Opus 4.6 measured |
| LMArena Text: English | Not measured | 1511 | Only Claude Opus 4.6 measured |
| LMArena Text: non-English | Not measured | 1488 | Only Claude Opus 4.6 measured |
| LMArena Text: Chinese | Not measured | 1551 | Only Claude Opus 4.6 measured |
| LMArena Text: German | Not measured | 1500 | Only Claude Opus 4.6 measured |
| LMArena Text: Japanese | Not measured | 1488 | Only Claude Opus 4.6 measured |
| LMArena Text: Korean | Not measured | 1464 | Only Claude Opus 4.6 measured |
| LMArena Text: French | Not measured | 1512 | Only Claude Opus 4.6 measured |
| LMArena Text: Spanish | Not measured | 1510 | Only Claude Opus 4.6 measured |
| LMArena Text: Russian | Not measured | 1497 | Only Claude Opus 4.6 measured |
| LMArena Text: Polish | Not measured | 1499 | Only Claude Opus 4.6 measured |
| LMArena Text: multi-turn | Not measured | 1511 | Only Claude Opus 4.6 measured |
| LMArena Text: longer queries | Not measured | 1518 | Only Claude Opus 4.6 measured |
| LMArena Text: writing and language | Not measured | 1502 | Only Claude Opus 4.6 measured |
| LMArena Text: entertainment and media | Not measured | 1492 | Only Claude Opus 4.6 measured |
| LMArena Vision: English | Not measured | 1324 | Only Claude Opus 4.6 measured |
| LMArena Vision: Chinese | Not measured | 1370 | Only Claude Opus 4.6 measured |
| LMArena Vision: reading text | Not measured | 1326 | Only Claude Opus 4.6 measured |
| LMArena Vision: humor | Not measured | 1314 | Only Claude Opus 4.6 measured |
| LMArena Vision: recognizing things | Not measured | 1274 | Only Claude Opus 4.6 measured |
| LMArena Vision: creative writing | Not measured | 1331 | Only Claude Opus 4.6 measured |
| LMArena Document | Not measured | 1507 | Only Claude Opus 4.6 measured |
| Neither model has a result on the other 9 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Claude Haiku 5.5 | Claude Opus 4.6 | Leader |
|---|---|---|---|
| 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 5.5$0.0003Fits in one request
Claude Opus 4.6$0.018Fits in one request
Claude Haiku 5.5 costs less
Repository review
50K input + 3K output tokens
Claude Haiku 5.5$0.0065Fits in one request
Claude Opus 4.6$0.325Fits in one request
Claude Haiku 5.5 costs less
Agent loop
200K input + 10K output tokens
Claude Haiku 5.5$0.025Fits in one request
Claude Opus 4.6$1.25Fits in one request
Claude Haiku 5.5 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.
| Specification | Anthropic Claude Haiku 5.5 | Anthropic Claude Opus 4.6 |
|---|---|---|
| Maker | Anthropic | Anthropic |
| Weights | No public weights listed | No public weights listed |
| Released | October 7, 2026 | February 4, 2026 |
| Context window | 1M | 1M |
| Input $/Mtok | $0.10 | $5.00 |
| Output $/Mtok | $0.50 | $25.00 |
| Coding index | 70.7 | 57.2 |
| Independent measurements behind it | 5 | 4 |
| Cost per solved task | Not computable | $1.56 |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Claude Haiku 5.5 better than Claude Opus 4.6 for coding?
Pick Claude Haiku 5.5 for coding: it scores 70.7 on the Vibe Coding Index against 57.2 for Claude Opus 4.6. Pick Claude Opus 4.6 for work measured by Vending-Bench 2 or GPQA Diamond (Epoch AI run), where it leads.
Which is cheaper, Claude Haiku 5.5 or Claude Opus 4.6?
Claude Haiku 5.5 costs $0.10 per million input tokens and $0.50 per million output tokens; Claude Opus 4.6 costs $5.00 and $25.00. Cost per solved task is known for Claude Opus 4.6 only.
Which has the bigger context window, Claude Haiku 5.5 or Claude Opus 4.6?
Neither: both take 1M tokens.
How many benchmarks is the Claude Haiku 5.5 vs Claude Opus 4.6 comparison based on?
34 benchmarks have a result for both models as of October 22, 2026. Claude Haiku 5.5's coding index rests on 5 independent benchmarks and Claude Opus 4.6's on 4. Where only one model has a result, the page says so instead of naming a winner.