Claude Opus 4.7 vs Claude Sonnet 4.5
Claude Opus 4.7 leads Claude Sonnet 4.5 on coding, 67.5 to 39.5 on the Vibe Coding Index, and Claude Sonnet 4.5 costs 1.7x less per output token ($15.00 vs $25.00 per million). Updated October 22, 2026; 71 benchmarks have a result for both.
Claude Opus 4.7 or Claude Sonnet 4.5: which to pick
Pick Claude Opus 4.7 for coding: it scores 67.5 on the Vibe Coding Index against 39.5 for Claude Sonnet 4.5.
Pick Claude Sonnet 4.5 for output tokens 1.7x cheaper ($15.00 vs $25.00 per million) and a lower cost per solved task ($1.16 against $2.95).
- Shared results
- 71
- Claude Opus 4.7 only
- 32
- Claude Sonnet 4.5 only
- 24
- Like-for-like categories
- 7 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 Opus 4.7
Claude Opus 4.7 scores 67.5 against 39.5.
4 and 3 benchmarks behind the scores
Intelligence
AA Intelligence Index
Claude Opus 4.7
Claude Opus 4.7 leads, 40.7 against 20.7.
Same board, both measured
Agentic work
AA Agentic Index
Claude Opus 4.7
Claude Opus 4.7 leads, 38.6 against 15.8.
Same board, both measured
Reasoning
GPQA Diamond
Claude Opus 4.7
Claude Opus 4.7 leads, 90.2% against 81.6%.
Same board, both measured
Math
AIME
Claude Opus 4.7
Claude Opus 4.7 leads, 96.3% against 88.2%.
Same board, both measured
Knowledge
MMLU Pro
Claude Opus 4.7
Claude Opus 4.7 leads, 89.9% against 87.4%.
Same board, both measured
Web design
Design Arena: websites
No call
Only Claude Sonnet 4.5 has a Design Arena: websites result, 1189.
Directional only
Long documents
Listed context window
No call
Both take 1M tokens.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Claude Sonnet 4.5
Claude Opus 4.7 $0.018, Claude Sonnet 4.5 $0.011 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
Claude Sonnet 4.5
Claude Opus 4.7 $2.95, Claude Sonnet 4.5 $1.16.
Published attempt cost and pass rate
At a glance
- Coding index
- Claude Opus 4.7 67.5 from 4 benchmarks, Claude Sonnet 4.5 39.5 from 3 benchmarks.
- Widest category
- Agentic work, 33.6 points apart on average over 2 shared benchmarks.
- Widest single gap
- ARC-AGI-2: Claude Opus 4.7 75.8% against Claude Sonnet 4.5 13.6%, 62.2 points apart on a 0 to 100 scale.
- Output price
- Claude Opus 4.7 $25.00 against Claude Sonnet 4.5 $15.00 per million output tokens, 1.7x 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 Opus 4.7 | Bars | Claude Sonnet 4.5 | Ties |
|---|---|---|---|---|
| Coding | 12 | 0 | ||
| Intelligence | 2 | 0 | ||
| Agentic work | 2 | 0 | ||
| Reasoning | 13 | 0 | ||
| Math | 7 | 0 | ||
| Knowledge | 13 | 0 | ||
| General | 21 | 1 | ||
| All shared boards | 70 | 1 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index120
| Benchmark | Claude Opus 4.7 | Claude Sonnet 4.5 | Leader |
|---|---|---|---|
| SWE-bench Verified | Not measured | 74.8% | Only Claude Sonnet 4.5 measured |
| SWE-bench Multilingual | Not measured | 67.0% | Only Claude Sonnet 4.5 measured |
| LMArena WebDev | 1558 | 1394 | Claude Opus 4.7 |
| AA Coding Index | 73.6 | 52.1 | Claude Opus 4.7 |
| LiveCodeBench (Vals AI) | 85.1% | 73.0% | Claude Opus 4.7 |
| SWE-bench Verified (Vals AI) | 82.0% | 70.0% | Claude Opus 4.7 |
| Vibe Code Bench (Vals AI) | 71.0% | 22.6% | Claude Opus 4.7 |
| Code migration (Vals AI) | 43.9% | Not measured | Only Claude Opus 4.7 measured |
| ProgramBench (Vals AI) | 0.0% | Not measured | Only Claude Opus 4.7 measured |
| Terminal-Bench 2.1 (Vals AI) | 68.5% | Not measured | Only Claude Opus 4.7 measured |
| Terminal-Bench Hard (Artificial Analysis run) | 54.5% | Not measured | Only Claude Opus 4.7 measured |
| Terminal-Bench | 80.2% | 46.5% | Claude Opus 4.7 |
| FrontierCode | 38.5% | Not measured | Only Claude Opus 4.7 measured |
| GSO-Bench | 44.1% | 14.7% | Claude Opus 4.7 |
| WeirdML | 76.4% | 47.7% | Claude Opus 4.7 |
| MirrorCode | 31.1% | Not measured | Only Claude Opus 4.7 measured |
| Cybench | Not measured | 60.0% | Only Claude Sonnet 4.5 measured |
| LMArena Text: coding | 1519 | 1488 | Claude Opus 4.7 |
| LMArena Text: software and IT | 1516 | 1473 | Claude Opus 4.7 |
| LMArena WebDev: React | 1560 | 1386 | Claude Opus 4.7 |
| LMArena WebDev: HTML | 1554 | 1399 | Claude Opus 4.7 |
| LMArena WebDev: from an image | 1577 | Not measured | Only Claude Opus 4.7 measured |
| Neither model has a result on the other 17 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | Claude Opus 4.7 | Claude Sonnet 4.5 | Leader |
|---|---|---|---|
| AA Intelligence Index | 40.7 | 20.7 | Claude Opus 4.7 |
| Epoch Capabilities Index | 156.8 | 146.7 | Claude Opus 4.7 |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | Claude Opus 4.7 | Claude Sonnet 4.5 | Leader |
|---|---|---|---|
| AA Agentic Index | 38.6 | 15.8 | Claude Opus 4.7 |
| Design Arena: agents building web apps | Not measured | 1060 | Only Claude Sonnet 4.5 measured |
| Design Arena: agents building full-stack apps | Not measured | 1050 | Only Claude Sonnet 4.5 measured |
| Design Arena: agents building mobile apps | Not measured | 1137 | Only Claude Sonnet 4.5 measured |
| Finance Agent (Vals AI) | 51.5% | Not measured | Only Claude Opus 4.7 measured |
| Tax Agent Bench (Vals AI) | 23.7% | Not measured | Only Claude Opus 4.7 measured |
| Legal Research Bench (Vals AI) | 38.5% | Not measured | Only Claude Opus 4.7 measured |
| Excel Modeling Benchmark (Vals AI) | 63.7% | Not measured | Only Claude Opus 4.7 measured |
| τ²-Bench Telecom | 88.6% | Not measured | Only Claude Opus 4.7 measured |
| APEX-Agents | 49.2% | Not measured | Only Claude Opus 4.7 measured |
| METR time horizons | Not measured | 67.4% | Only Claude Sonnet 4.5 measured |
| Remote Labor Index | Not measured | 2.1% | Only Claude Sonnet 4.5 measured |
| GDPval | Not measured | 42.5% | Only Claude Sonnet 4.5 measured |
| OSWorld 2.0 | 18.2% | Not measured | Only Claude Opus 4.7 measured |
| DeepResearch Bench | Not measured | 52.6% | Only Claude Sonnet 4.5 measured |
| PostTrainBench | 28.6% | Not measured | Only Claude Opus 4.7 measured |
| OSWorld | Not measured | 62.9% | Only Claude Sonnet 4.5 measured |
| Vending-Bench 2 | $10,937 | $3,839 | Claude Opus 4.7 |
| LMArena Search | 1230 | Not measured | Only Claude Opus 4.7 measured |
| Berkeley Function Calling Leaderboard (BFCL) | Not measured | 73.2% | Only Claude Sonnet 4.5 measured |
| Neither model has a result on the other 21 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index130
| Benchmark | Claude Opus 4.7 | Claude Sonnet 4.5 | Leader |
|---|---|---|---|
| GPQA Diamond | 90.2% | 81.6% | Claude Opus 4.7 |
| MMMU Pro (Vals AI) | 85.5% | 79.3% | Claude Opus 4.7 |
| GPQA Diamond (Artificial Analysis run) | 91.4% | Not measured | Only Claude Opus 4.7 measured |
| Humanity's Last Exam (Artificial Analysis run) | 42.3% | Not measured | Only Claude Opus 4.7 measured |
| GPQA Diamond (Epoch AI run) | 90.2% | 82.3% | Claude Opus 4.7 |
| ARC-AGI | 93.5% | 63.7% | Claude Opus 4.7 |
| ARC-AGI-2 | 75.8% | 13.6% | Claude Opus 4.7 |
| SimpleBench | 61.7% | 54.3% | Claude Opus 4.7 |
| Chess puzzles | 41.0% | 10.0% | Claude Opus 4.7 |
| Mystery game puzzles | 28.0% | 17.0% | Claude Opus 4.7 |
| EBR-bench | 19.0% | 2.4% | Claude Opus 4.7 |
| VPCT | Not measured | 39.8% | Only Claude Sonnet 4.5 measured |
| LMArena Text: hard prompts | 1506 | 1462 | Claude Opus 4.7 |
| LMArena Text: expert | 1521 | 1483 | Claude Opus 4.7 |
| LMArena Text: hard prompts in English | 1508 | 1471 | Claude Opus 4.7 |
| LMArena Vision: diagrams | 1336 | Not measured | Only Claude Opus 4.7 measured |
| LMArena Vision: homework | 1337 | Not measured | Only Claude Opus 4.7 measured |
| Kagi LLM Benchmark | 80.7% | 57.9% | Claude Opus 4.7 |
| Neither model has a result on the other 4 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index70
| Benchmark | Claude Opus 4.7 | Claude Sonnet 4.5 | Leader |
|---|---|---|---|
| AIME | 96.3% | 88.2% | Claude Opus 4.7 |
| MGSM (Vals AI) | Not measured | 94.3% | Only Claude Sonnet 4.5 measured |
| OTIS Mock AIME | 97.8% | 77.8% | Claude Opus 4.7 |
| FrontierMath, tiers 1 to 3 | 70.2% | 23.9% | Claude Opus 4.7 |
| FrontierMath, tier 4 | 31.7% | 2.4% | Claude Opus 4.7 |
| ProofBench | 54.0% | 19.0% | Claude Opus 4.7 |
| MATH level 5 | Not measured | 97.7% | Only Claude Sonnet 4.5 measured |
| LMArena Text: math | 1498 | 1449 | Claude Opus 4.7 |
| LMArena Text: mathematical work | 1509 | 1462 | Claude Opus 4.7 |
| Neither model has a result on the other 4 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index130
| Benchmark | Claude Opus 4.7 | Claude Sonnet 4.5 | Leader |
|---|---|---|---|
| MMLU Pro | 89.9% | 87.4% | Claude Opus 4.7 |
| LegalBench (Vals AI) | 85.3% | 84.1% | Claude Opus 4.7 |
| CaseLaw (Vals AI) | 68.4% | 62.2% | Claude Opus 4.7 |
| CorpFin (Vals AI) | 66.1% | 62.0% | Claude Opus 4.7 |
| TaxEval (Vals AI) | 75.3% | 73.3% | Claude Opus 4.7 |
| MortgageTax (Vals AI) | 70.3% | 64.0% | Claude Opus 4.7 |
| MedQA (Vals AI) | Not measured | 94.7% | Only Claude Sonnet 4.5 measured |
| MedCode (Vals AI) | 54.9% | 44.1% | Claude Opus 4.7 |
| SimpleQA Verified | 51.7% | 30.7% | Claude Opus 4.7 |
| LMArena Text: science | 1508 | 1448 | Claude Opus 4.7 |
| LMArena Text: medicine and healthcare | 1495 | 1444 | Claude Opus 4.7 |
| LMArena Text: legal and government | 1504 | 1451 | Claude Opus 4.7 |
| LMArena Text: business and finance | 1490 | 1440 | Claude Opus 4.7 |
| FACTS Benchmark Suite | 45.5% | 40.6% | Claude Opus 4.7 |
| Neither model has a result on the other 4 boards here | |||
Web designDirectional onlyPublic boards, shown as published, kept out of the index8 boards, none shared
| Benchmark | Claude Opus 4.7 | Claude Sonnet 4.5 | Leader |
|---|---|---|---|
| Design Arena: websites | Not measured | 1189 | Only Claude Sonnet 4.5 measured |
| Design Arena: UI components | Not measured | 1170 | Only Claude Sonnet 4.5 measured |
| Design Arena: game dev | Not measured | 1166 | Only Claude Sonnet 4.5 measured |
| Design Arena: data viz | Not measured | 1167 | Only Claude Sonnet 4.5 measured |
| Design Arena: 3D | Not measured | 1157 | Only Claude Sonnet 4.5 measured |
| Design Arena: all code categories | Not measured | 1182 | Only Claude Sonnet 4.5 measured |
| Design Arena: SVG | Not measured | 1115 | Only Claude Sonnet 4.5 measured |
| Design Arena: ASCII art | Not measured | 1205 | Only Claude Sonnet 4.5 measured |
| Neither model has a result on the other 1 board here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index211
| Benchmark | Claude Opus 4.7 | Claude Sonnet 4.5 | Leader |
|---|---|---|---|
| SAGE (Vals AI) | 56.1% | 36.1% | Claude Opus 4.7 |
| MedScribe (Vals AI) | 83.0% | 84.5% | Claude Sonnet 4.5 |
| Vals Multimodal Index | 67.4% | Not measured | Only Claude Opus 4.7 measured |
| IFBench | 58.6% | Not measured | Only Claude Opus 4.7 measured |
| Long context reasoning (AA-LCR) | 78.7% | Not measured | Only Claude Opus 4.7 measured |
| DTBench | 94.7% | 83.2% | Claude Opus 4.7 |
| LMCA | 52.2% | 38.8% | Claude Opus 4.7 |
| Furniture assembly | 33.3% | Not measured | Only Claude Opus 4.7 measured |
| ExploitBench | 26.5% | Not measured | Only Claude Opus 4.7 measured |
| LMArena Text | 1490 | 1438 | Claude Opus 4.7 |
| LMArena Text: instruction following | 1498 | 1459 | Claude Opus 4.7 |
| LMArena Text: creative writing | 1486 | 1442 | Claude Opus 4.7 |
| LMArena Vision | 1317 | Not measured | Only Claude Opus 4.7 measured |
| LMArena Text: English | 1496 | 1450 | Claude Opus 4.7 |
| LMArena Text: non-English | 1480 | 1425 | Claude Opus 4.7 |
| LMArena Text: Chinese | 1531 | 1460 | Claude Opus 4.7 |
| LMArena Text: German | 1495 | 1427 | Claude Opus 4.7 |
| LMArena Text: Japanese | 1474 | 1390 | Claude Opus 4.7 |
| LMArena Text: Korean | 1464 | 1402 | Claude Opus 4.7 |
| LMArena Text: French | 1503 | 1459 | Claude Opus 4.7 |
| LMArena Text: Spanish | 1494 | 1458 | Claude Opus 4.7 |
| LMArena Text: Russian | 1494 | 1437 | Claude Opus 4.7 |
| LMArena Text: Polish | 1489 | 1423 | Claude Opus 4.7 |
| LMArena Text: multi-turn | 1505 | 1465 | Claude Opus 4.7 |
| LMArena Text: longer queries | 1505 | 1475 | Claude Opus 4.7 |
| LMArena Text: writing and language | 1492 | 1442 | Claude Opus 4.7 |
| LMArena Text: entertainment and media | 1474 | 1428 | Claude Opus 4.7 |
| LMArena Vision: English | 1322 | Not measured | Only Claude Opus 4.7 measured |
| LMArena Vision: Chinese | 1371 | Not measured | Only Claude Opus 4.7 measured |
| LMArena Vision: reading text | 1327 | Not measured | Only Claude Opus 4.7 measured |
| LMArena Vision: humor | 1309 | Not measured | Only Claude Opus 4.7 measured |
| LMArena Vision: recognizing things | 1278 | Not measured | Only Claude Opus 4.7 measured |
| LMArena Vision: creative writing | 1326 | Not measured | Only Claude Opus 4.7 measured |
| LMArena Document | 1495 | 1450 | Claude Opus 4.7 |
| 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
| Benchmark | Claude Opus 4.7 | Claude Sonnet 4.5 | 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 Sonnet 4.5 costs less
Repository review
50K input + 3K output tokens
Claude Sonnet 4.5 costs less
Agent loop
200K input + 10K output tokens
Claude Sonnet 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.
| Specification | Anthropic Claude Opus 4.7 | Anthropic Claude Sonnet 4.5 |
|---|---|---|
| Maker | Anthropic | Anthropic |
| Weights | No public weights listed | No public weights listed |
| Released | April 16, 2026 | September 29, 2025 |
| Context window | 1M | 1M |
| Input $/Mtok | $5.00 | $3.00 |
| Output $/Mtok | $25.00 | $15.00 |
| Coding index | 67.5 | 39.5 |
| Independent measurements behind it | 4 | 3 |
| Cost per solved task | $2.95 | $1.16 |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Claude Opus 4.7 better than Claude Sonnet 4.5 for coding?
Pick Claude Opus 4.7 for coding: it scores 67.5 on the Vibe Coding Index against 39.5 for Claude Sonnet 4.5. Pick Claude Sonnet 4.5 for output tokens 1.7x cheaper ($15.00 vs $25.00 per million) and a lower cost per solved task ($1.16 against $2.95).
Which is cheaper, Claude Opus 4.7 or Claude Sonnet 4.5?
Claude Opus 4.7 costs $5.00 per million input tokens and $25.00 per million output tokens; Claude Sonnet 4.5 costs $3.00 and $15.00. Per solved SWE-bench Verified task, Claude Opus 4.7 costs $2.95 and Claude Sonnet 4.5 $1.16.
Which has the bigger context window, Claude Opus 4.7 or Claude Sonnet 4.5?
Neither: both take 1M tokens.
How many benchmarks is the Claude Opus 4.7 vs Claude Sonnet 4.5 comparison based on?
71 benchmarks have a result for both models as of October 22, 2026. Claude Opus 4.7's coding index rests on 4 independent benchmarks and Claude Sonnet 4.5's on 3. Where only one model has a result, the page says so instead of naming a winner.