Sign Up

Claude Opus 4.8

Anthropic
Announcement
Claude Opus 4.8 launch post cover

Cost/1M tokens

input
$5.00
cached
$0.50
output
$25.00
cache write
$6.25
Context
1Mtokens
Released
5mo 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

#15 of 28

  1. 74.1%
    GPT-6 Astra
  2. 73.8%
    Gemini 3.8 Flash
  3. 73.6%
    Claude Opus 5
  4. 72.7%
    GPT-5.6 Sol
  5. 69.9%
    Claude Fable 5
  6. 69.6%
    GPT-5.6 Terra
  7. 69.0%
    GLM 5.3
  8. 68.5%
    Kimi K3
  9. 67.5%
    Grok 4.6
  10. 67.2%
    GPT-5.6 Luna
  11. 59.0%
    Claude Opus 4.8

Arena Text

Best Arena rating per model · Higher is better

Arena #28

  1. 1511
    Muse WonderSimilar price
  2. 1501
    Claude Opus 4.7Predecessor
  3. 1485
    GPT-5.6 SolSimilar price
  4. 1483
    GLM 5.3Nearby rank
  5. 1482
    GPT-6.1 SolNearby rank
  6. 1482
    Claude Opus 4.8
  7. 1481
    GPT-5.5Similar price
  8. 1480
    MiMo-V2.6-ProNearby rank
  9. 1346
    GPT-4o (2024-05-13)Similar price

-19 vs Claude Opus 4.7 (predecessor) · Ahead of 1 of 4 similarly priced models

Arena WebDev

Best Arena rating per model · Higher is better

Arena #49

  1. 1649
    Muse WonderSimilar price
  2. 1615
    GPT-5.6 SolSimilar price
  3. 1564
    Mistral Large 4Nearby rank
  4. 1557
    Claude Opus 4.7Predecessor
  5. 1555
    Claude Opus 4.8
  6. 1552
    Grok 4.5Nearby rank
  7. 1547
    Claude Opus 4.6Nearby rank
  8. 1512
    GPT-5.5Similar price
  9. 1462
    GPT-5.4Similar price

Level with Claude Opus 4.7 (predecessor, overlapping 95% CIs) · Ahead of 2 of 4 similarly priced models

Arena Vision

Best Arena rating per model · Higher is better

Arena #18

  1. 1300
    Claude Opus 4.7Predecessor
  2. 1289
    Claude Fable 5.1Nearby rank
  3. 1288
    GPT-5.6 SolSimilar price
  4. 1286
    Claude Opus 4.8
  5. 1285
    Claude Opus 5Nearby rank
  6. 1285
    GPT-5.5Similar price
  7. 1282
    GPT-5.4Similar price
  8. 1162
    GPT-4o (2024-05-13)Similar price

-14 vs Claude Opus 4.7 (predecessor) · Ahead of 1 of 4 similarly priced models

Arena Search

Best Arena rating per model · Higher is better

Arena #23

  1. 1255
    GPT-5.6 SolSimilar price
  2. 1245
    GPT-6 AstraSimilar price
  3. 1213
    Claude Opus 5Similar price
  4. 1210
    Claude Opus 4.7Predecessor
  5. 1205
    Grok 4.5Nearby rank
  6. 1201
    Gemini 3 Pro PreviewNearby rank
  7. 1194
    Claude Opus 4.8
  8. 1194
    Grok 4.3Nearby rank
  9. 1192
    ERNIE 5.1Nearby rank
  10. 1142
    Sonar ProSimilar price

-16 vs Claude Opus 4.7 (predecessor) · Ahead of 1 of 4 similarly priced models

Arena Document

Best Arena rating per model · Higher is better

Arena #9

  1. 1494
    Claude Opus 4.7Predecessor
  2. 1486
    GPT-5.6 SolSimilar price
  3. 1486
    Claude Opus 5Nearby rank
  4. 1485
    Claude Opus 4.8
  5. 1481
    GPT-5.5Similar price
  6. 1480
    Muse Spark 1.1Nearby rank
  7. 1474
    GPT-6 AstraSimilar price
  8. 1471
    GPT-5.4Similar price

Level with Claude Opus 4.7 (predecessor, overlapping 95% CIs) · Ahead of 1 of 4 similarly priced models

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

#25 of 85

  1. 95.0%
    GPT-6 AstraSimilar price
  2. 92.5%
    GPT-5.6 SolSimilar price
  3. 85.0%
    GPT-5.5Similar price
  4. 75.8%
    Claude Opus 4.7Predecessor
  5. 74.0%
    GPT-5.4Similar price
  6. 72.9%
    DeepSeek V4.1 FlashNearby rank
  7. 72.1%
    Claude Opus 4.8
  8. 72.1%
    Gemini 3.5 FlashNearby rank
  9. 67.1%
    Grok 4.6Nearby rank

-3.7 pts vs Claude Opus 4.7 (predecessor) · Ahead of 0 of 4 similarly priced models

FrontierMath-2025-02-28-Private

Best score per model · Higher is better

#3 of 58

  1. 50.0%
    GPT-5.4 ProNearby rank
  2. 47.6%
    GPT-5.4Similar price
  3. 47.2%
    Claude Opus 4.8
  4. 43.8%
    Claude Opus 4.7Predecessor
  5. 40.7%
    Claude Opus 4.6Similar price
  6. 32.4%
    Claude Sonnet 4.6Similar price
  7. 20.7%
    Claude Opus 4.5Similar price

+3.4 pts vs Claude Opus 4.7 (predecessor) · Ahead of 3 of 4 similarly priced models

FrontierMath-Tier-4-2025-07-01-Private

Best score per model · Higher is better

#1 of 46

  1. 31.3%
    Claude Opus 4.8
  2. 27.1%
    GPT-5.4Similar price
  3. 22.9%
    Claude Opus 4.7Predecessor
  4. 22.9%
    Claude Opus 4.6Similar price
  5. 8.3%
    Claude Sonnet 4.6Similar price
  6. 4.2%
    Claude Opus 4.5Similar price

+8.3 pts vs Claude Opus 4.7 (predecessor) · Ahead of 4 of 4 similarly priced models

FrontierMath-Tier-4-v2-Private

Best score per model · Higher is better

#20 of 72

  1. 97.6%
    GPT-6 AstraSimilar price
  2. 82.9%
    GPT-5.6 SolSimilar price
  3. 72.5%
    GPT-5.5Similar price
  4. 61.0%
    GPT-5.6 LunaNearby rank
  5. 58.5%
    GPT-5.4 ProNearby rank
  6. 56.1%
    Claude Opus 4.8
  7. 56.1%
    GPT-6 LunaNearby rank
  8. 49.0%
    GPT-5.4Similar price
  9. 31.7%
    Claude Opus 4.7Predecessor

+24.4 pts vs Claude Opus 4.7 (predecessor) · Ahead of 1 of 4 similarly priced models

FrontierMath-Tiers-1-3-v2-Private

Best score per model · Higher is better

#19 of 87

  1. 93.7%
    GPT-6 AstraSimilar price
  2. 89.1%
    GPT-5.6 SolSimilar price
  3. 85.3%
    GPT-5.5Similar price
  4. 82.5%
    GPT-5.4 ProNearby rank
  5. 82.1%
    GPT-5.6 LunaNearby rank
  6. 80.0%
    Claude Opus 4.8
  7. 78.9%
    GPT-6 LunaNearby rank
  8. 78.6%
    GPT-5.4Similar price
  9. 70.2%
    Claude Opus 4.7Predecessor

+9.8 pts vs Claude Opus 4.7 (predecessor) · Ahead of 1 of 4 similarly priced models

GPQA diamond

Best score per model · Higher is better

#31 of 161

  1. 93.5%
    GPT-5.6 SolSimilar price
  2. 93.3%
    GPT-5.4Similar price
  3. 91.6%
    GPT-5.6 LunaNearby rank
  4. 91.4%
    GPT-5.2Nearby rank
  5. 91.0%
    Claude Opus 4.8
  6. 91.0%
    DeepSeek V4 Flash 0731Nearby rank
  7. 90.9%
    DeepSeek V4 Pro 0423Nearby rank
  8. 90.7%
    GPT-5.5Similar price
  9. 90.2%
    Claude Opus 4.7Predecessor
  10. 48.9%
    GPT-4o (2024-05-13)Similar price

+0.9 pts vs Claude Opus 4.7 (predecessor) · Ahead of 2 of 4 similarly priced models

OSWorld 2.0

Best score per model · Higher is better

#3 of 9

  1. 31.4%
    Claude Opus 5Similar price
  2. 27.3%
    GPT-5.6 SolSimilar price
  3. 18.5%
    Claude Opus 4.8
  4. 13.9%
    Claude Opus 4.7Predecessor
  5. 13.0%
    GPT-5.5Similar price
  6. 9.3%
    Claude Sonnet 4.6Similar price

+4.6 pts vs Claude Opus 4.7 (predecessor) · Ahead of 2 of 4 similarly priced models

OTIS Mock AIME 2024-2025

Best score per model · Higher is better

#24 of 158

  1. 100.0%
    GPT-5.6 SolSimilar price
  2. 98.9%
    GPT-6 LunaNearby rank
  3. 98.6%
    DeepSeek V4 Pro 0813Nearby rank
  4. 98.3%
    Claude Opus 4.8
  5. 98.3%
    DeepSeek V4.1 FlashNearby rank
  6. 98.3%
    Mistral Large 4Nearby rank
  7. 97.8%
    Claude Opus 4.7Predecessor
  8. 97.8%
    GPT-5.4Similar price
  9. 84.4%
    GPT-5.5Similar price
  10. 6.3%
    GPT-4o (2024-05-13)Similar price

+0.5 pts vs Claude Opus 4.7 (predecessor) · Ahead of 3 of 4 similarly priced models

SimpleQA Verified

Best score per model · Higher is better

#24 of 87

  1. 75.6%
    GPT-6 AstraSimilar price
  2. 69.7%
    GPT-5.6 SolSimilar price
  3. 63.0%
    GPT-5.5Similar price
  4. 56.0%
    Grok 4.7Nearby rank
  5. 55.8%
    Qwen3.7 MaxNearby rank
  6. 53.0%
    Claude Opus 4.8
  7. 52.9%
    DeepSeek V4 Pro 0813Nearby rank
  8. 52.0%
    Qwen3.6 Max PreviewNearby rank
  9. 51.7%
    Claude Opus 4.7Predecessor
  10. 45.1%
    GPT-5.4Similar price

+1.3 pts vs Claude Opus 4.7 (predecessor) · Ahead of 1 of 4 similarly priced models

DeepSWE 1.1

Primary coding-agent benchmark · 113 tasks · Data generated 2026-09-22

Official leaderboard
59.0%
pass@1 · highest published configuration
mini-swe-agent · max reasoning
pass@4
79.3%
Average cost / attempt
$13.22
Benchmark runs
4

Score vs cost by reasoning effort

Average cost per attempt (USD) against pass rate

All published configurations

Harness / reasoningpass@1pass@4Cost / attempt
mini-swe-agent · max59.0%79.3%$13.22
mini-swe-agent · xhigh54.4%80.5%$8.01
mini-swe-agent · high51.8%77.9%$4.28
mini-swe-agent · medium48.7%76.1%$3.44
mini-swe-agent · low40.8%68.1%$2.29

pass@1 is the success rate across scored attempts; pass@4 is the share of tasks solved in at least one of four attempts. Costs are reported by DeepSWE for each configuration. These results are not directly comparable to evaluations using different harnesses. Official source JSON.

Reasoning effort

ARC-AGI-2

Score by reasoning effort · Higher is better

GPQA diamond

Score by reasoning effort · Higher is better

OTIS Mock AIME 2024-2025

Score by reasoning effort · Higher is better

All published scores

  • Arena TextArena #281482

    -19 vs Claude Opus 4.7 (predecessor)

    • high148295% CI 1478 to 1486 · 63,128 votes
    • default147495% CI 1470 to 1478 · 64,170 votes
    Source:Arena Text leaderboard, 2026-11-09
  • Arena WebDevArena #491555

    Level with Claude Opus 4.7 (predecessor, overlapping 95% CIs)

    • high155595% CI 1550 to 1561 · 18,030 votes
    • default153695% CI 1531 to 1542 · 16,966 votes
    Source:Arena WebDev leaderboard, 2026-11-10
  • Arena VisionArena #181286

    -14 vs Claude Opus 4.7 (predecessor)

    • high128695% CI 1279 to 1292 · 18,178 votes
    • default127995% CI 1272 to 1285 · 18,692 votes
    Source:Arena Vision leaderboard, 2026-11-09
  • Arena SearchArena #231194

    -16 vs Claude Opus 4.7 (predecessor)

    • default119495% CI 1188 to 1200 · 101,197 votes
    Source:Arena Search leaderboard, 2026-11-09
  • Arena DocumentArena #91485

    Level with Claude Opus 4.7 (predecessor, overlapping 95% CIs)

    • high148595% CI 1479 to 1492 · 12,069 votes
    • default147695% CI 1470 to 1482 · 14,729 votes
    Source:Arena Document leaderboard, 2026-11-09
  • ARC-AGI-2#25 of 8572.1%

    -3.7 pts vs Claude Opus 4.7 (predecessor)

    • low62.2%
    • medium71.7%
    • high72.1%
    Source:External evaluation via Epoch AIOriginal-source terms apply
  • FrontierMath-2025-02-28-Private#3 of 5847.2%

    +3.4 pts vs Claude Opus 4.7 (predecessor)

    • max47.2%
    Source:Epoch AI evaluation
  • FrontierMath-Tier-4-2025-07-01-Private#1 of 4631.3%

    +8.3 pts vs Claude Opus 4.7 (predecessor)

    • max31.3%
    Source:Epoch AI evaluation
  • FrontierMath-Tier-4-v2-Private#20 of 7256.1%

    +24.4 pts vs Claude Opus 4.7 (predecessor)

    • max56.1%
    Source:Epoch AI evaluation
  • FrontierMath-Tiers-1-3-v2-Private#19 of 8780.0%

    +9.8 pts vs Claude Opus 4.7 (predecessor)

    • max80.0%
    Source:Epoch AI evaluation
  • GPQA diamond#31 of 16191.0%

    +0.9 pts vs Claude Opus 4.7 (predecessor)

    • none85.4%
    • low88.4%
    • max91.0%
    Source:Epoch AI evaluation
  • OSWorld 2.0#3 of 918.5%

    +4.6 pts vs Claude Opus 4.7 (predecessor)

    • max18.5%
    Source:https://osworld-v2.xlang.ai/Original-source terms apply
  • OTIS Mock AIME 2024-2025#24 of 15898.3%

    +0.5 pts vs Claude Opus 4.7 (predecessor)

    • none84.4%
    • low97.8%
    • max98.3%
    Source:Epoch AI evaluation
  • SimpleQA Verified#24 of 8753.0%

    +1.3 pts vs Claude Opus 4.7 (predecessor)

    • max53.0%
    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.