Sign Up

Claude Sonnet 3.7

Anthropic
Announcement
Claude Sonnet 3.7 launch post cover
Context
200Ktokens
Released
1y ago

Benchmarks

DeepSWE is the primary coding benchmark where an exact model evaluation exists. Arena leaderboards and other sourced evaluations appear below; missing scores are not zero.

How it ranks

Against its predecessor, similarly priced models and the models ranked around it.

DeepSWE 1.1

Best pass@1 per model · Higher is better

Not evaluated in DeepSWE 1.1

Arena Text

Best Arena rating per model · Higher is better

Arena #184

  1. 1391
    MiniMax M2.5Nearby rank
  2. 1390
    GPT-5 MiniNearby rank
  3. 1388
    Claude Sonnet 3.7
  4. 1387
    Solar Pro 4Nearby rank
  5. 1387
    Mistral Medium 3Nearby rank
  6. 1374
    Claude Sonnet 3.5 v2Predecessor
  7. 1365
    Gemma 3 27BSimilar release date
  8. 1354
    Command ASimilar release date
  9. 1336
    QwQ 32BSimilar release date
  10. 1274
    Mistral Small 3Similar release date

+14 vs Claude Sonnet 3.5 v2 (predecessor) · Ahead of 4 of 4 models released around the same time

Arena WebDev

Best Arena rating per model · Higher is better

Not ranked on Arena WebDev yet

Arena Vision

Best Arena rating per model · Higher is better

Arena #98

  1. 1199
    GPT-5.4 NanoNearby rank
  2. 1196
    Grok 4.1 Fast (Reasoning)Nearby rank
  3. 1193
    Claude Sonnet 3.7
  4. 1189
    Gemini 2.5 Flash-Lite Preview 06-17Nearby rank
  5. 1188
    Qwen3 VL 235B A22B ThinkingNearby rank
  6. 1161
    Claude Sonnet 3.5 v2Predecessor
  7. 1160
    Gemma 3 27BSimilar release date
  8. 1127
    Mistral Small 3.1 24BSimilar release date
  9. 1121
    Qwen2.5 VL 72B InstructSimilar release date
  10. 1001
    Aya Vision 32BSimilar release date

+31 vs Claude Sonnet 3.5 v2 (predecessor) · Ahead of 4 of 4 models released around the same time

Arena Search

Best Arena rating per model · Higher is better

Not ranked on Arena Search yet

Arena Document

Best Arena rating per model · Higher is better

Not ranked on Arena Document yet

Agent Arena

Best IPS score per model · Higher is better

Not ranked on Agent Arena yet

ARC-AGI-2

Best score per model · Higher is better

#76 of 85

  1. 2.5%
    Gemini 2.5 Flash Preview 05-20Similar release date
  2. 1.3%
    Qwen3 235B A22B Instruct 2507Nearby rank
  3. 1.1%
    R1 0528Similar release date
  4. 0.9%
    Claude Sonnet 3.7
  5. 0.4%
    GPT-4.1Similar release date
  6. 0.0%
    Gemini 2.5 Pro Preview 06-05Nearby rank
  7. 0.0%
    Llama 4 MaverickSimilar release date

Ahead of 2 of 4 models released around the same time

FrontierMath-2025-02-28-Private

Best score per model · Higher is better

#44 of 58

  1. 10.3%
    Gemini 2.5 Pro Preview 06-05Similar release date
  2. 5.5%
    GPT-4.1Similar release date
  3. 4.5%
    Claude Opus 4Nearby rank
  4. 4.5%
    GPT-4.1 MiniNearby rank
  5. 4.1%
    Claude Sonnet 3.7
  6. 4.1%
    Claude Sonnet 4Nearby rank
  7. 3.8%
    GLM 4.6Nearby rank
  8. 2.1%
    Claude Sonnet 3.5 v2Predecessor
  9. 0.3%
    Mistral Medium 3Similar release date
  10. 0.0%
    Llama 4 ScoutSimilar release date

+2.1 pts vs Claude Sonnet 3.5 v2 (predecessor) · Ahead of 2 of 4 models released around the same time

GPQA diamond

Best score per model · Higher is better

#88 of 161

  1. 81.8%
    o3Nearby rank
  2. 80.1%
    Qwen3 235B A22B Thinking 2507Nearby rank
  3. 79.7%
    Claude Sonnet 3.7
  4. 79.2%
    Claude Sonnet 4Nearby rank
  5. 79.0%
    Qwen3.5-9BNearby rank
  6. 65.4%
    QwQ PlusSimilar release date
  7. 64.1%
    DeepSeek-R1-Distill-Qwen-32BSimilar release date
  8. 55.3%
    Claude Sonnet 3.5 v2Predecessor
  9. 51.8%
    Llama 4 ScoutSimilar release date
  10. 47.7%
    Gemma 3 27BSimilar release date

+24.4 pts vs Claude Sonnet 3.5 v2 (predecessor) · Ahead of 4 of 4 models released around the same time

Humanity's Last Exam

Best score per model · Higher is better

#25 of 34

  1. 11.0%
    Gemini 2.5 Flash Preview 05-20Similar release date
  2. 8.3%
    GLM 4.5Nearby rank
  3. 8.1%
    GLM 4.5 AirNearby rank
  4. 8.0%
    Claude Sonnet 3.7
  5. 7.9%
    Claude Opus 4.1Nearby rank
  6. 7.5%
    Claude Sonnet 4.5Nearby rank
  7. 5.7%
    Llama 4 MaverickSimilar release date
  8. 5.4%
    GPT-4.1Similar release date
  9. 4.5%
    Mistral Medium 3Similar release date
  10. 4.1%
    Claude Sonnet 3.5 v2Predecessor

+4.0 pts vs Claude Sonnet 3.5 v2 (predecessor) · Ahead of 3 of 4 models released around the same time

OSWorld

Best score per model · Higher is better

#8 of 10

  1. 62.9%
    Claude Sonnet 4.5Nearby rank
  2. 43.9%
    Claude Sonnet 4Similar release date
  3. 35.8%
    Claude Sonnet 3.7
  4. 23.0%
    o3Similar release date
  5. 5.0%
    Qwen2.5 VL 72B InstructSimilar release date

Ahead of 2 of 3 models released around the same time

OTIS Mock AIME 2024-2025

Best score per model · Higher is better

#115 of 158

  1. 59.2%
    QwQ 32BSimilar release date
  2. 58.3%
    GLM 4.7 FlashNearby rank
  3. 57.8%
    Claude Sonnet 3.7
  4. 56.1%
    Qwen3 8BNearby rank
  5. 55.8%
    Qwen3.5 4BNearby rank
  6. 55.6%
    DeepSeek-R1-Distill-Qwen-32BSimilar release date
  7. 22.5%
    Gemma 3 27BSimilar release date
  8. 8.5%
    Claude Sonnet 3.5 v2Predecessor
  9. 7.8%
    Llama 4 ScoutSimilar release date

+49.3 pts vs Claude Sonnet 3.5 v2 (predecessor) · Ahead of 3 of 4 models released around the same time

SWE-bench Verified

Best score per model · Higher is better

#31 of 35

  1. 64.7%
    GPT-5 MiniNearby rank
  2. 62.3%
    o3Similar release date
  3. 61.0%
    Claude Sonnet 3.7
  4. 57.9%
    Qwen3.6 PlusNearby rank
  5. 57.6%
    Gemini 2.5 ProSimilar release date
  6. 48.5%
    GPT-4.1Similar release date
  7. 31.0%
    GPT-4o (2024-11-20)Similar release date

Ahead of 3 of 4 models released around the same time

Not evaluated in DeepSWE 1.1 as of 2026-09-22. Results are only shown for exact model versions in the official dataset. Official leaderboard.

All published scores

  • Arena TextArena #1841388

    +14 vs Claude Sonnet 3.5 v2 (predecessor)

    • default138832K, thinking95% CI 1384 to 1393 · 38,126 votes
    • default137295% CI 1368 to 1376 · 42,553 votes
    Source:Arena Text leaderboard, 2026-11-09
  • Arena VisionArena #981193

    +31 vs Claude Sonnet 3.5 v2 (predecessor)

    • default119332K, thinking95% CI 1177 to 1208 · 1,432 votes
    • default117595% CI 1166 to 1185 · 4,516 votes
    Source:Arena Vision leaderboard, 2026-11-09
  • ARC-AGI-2#76 of 850.9%
    • default0.9%8K
    • default0.7%16K
    • default0.4%1K
    • default0.0%
    Source:External evaluation via Epoch AIOriginal-source terms apply
  • FrontierMath-2025-02-28-Private#44 of 584.1%

    +2.1 pts vs Claude Sonnet 3.5 v2 (predecessor)

    • default4.1%16K
    • default3.4%32K
    • default3.1%
    • default3.1%64K
    Source:Epoch AI evaluation
  • GPQA diamond#88 of 16179.7%

    +24.4 pts vs Claude Sonnet 3.5 v2 (predecessor)

    • default79.7%64K
    • default76.8%16K
    • default76.8%32K
    • default66.0%
    Source:Epoch AI evaluation
  • Humanity's Last Exam#25 of 348.0%

    +4.0 pts vs Claude Sonnet 3.5 v2 (predecessor)

    • unknown8.0%
    Source:External evaluation via Epoch AIOriginal-source terms apply
  • OSWorld#8 of 1035.8%
    • default35.8%claude-3-7-sonnet-20250219 (50 steps)
    • default35.6%claude-3-7-sonnet-20250219 (100 steps)
    • default27.1%claude-3-7-sonnet-20250219 (15 steps)
    Source:OS World WebsiteOriginal-source terms apply
  • OTIS Mock AIME 2024-2025#115 of 15857.8%

    +49.3 pts vs Claude Sonnet 3.5 v2 (predecessor)

    • default57.8%64K
    • default53.3%32K
    • default46.7%16K
    • default21.9%
    Source:Epoch AI evaluation
  • SWE-bench Verified#31 of 3561.0%
    • default61.0%
    Source:Epoch AI evaluation

Benchmark compilation: Epoch AI, Capabilities & benchmarking (CC BY attribution). Scores are converted to percentages and model names matched to the Vertical Tabs catalog. Individual evaluations are credited to their original sources. External results may carry separate licences and evaluation configurations; do not interpret scores from different harnesses as directly comparable.

Arena ratings: Arena (lmarena-ai/leaderboard-dataset), licensed CC BY 4.0. Ratings come from pairwise human votes and are only comparable within one leaderboard; the interval is Arena's 95% confidence interval. Model names are matched to the Vertical Tabs catalog.

To give you the best experience, we use cookies for personalization, insights, and performance. Learn more.