Sign Up

Claude Haiku 4.5

Anthropic
Announcement
Claude Haiku 4.5 launch post cover

Cost/1M tokens

input
$1.00
cached
$0.10
output
$5.00
cache write
$1.25
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 #149

  1. 1496
    Muse Spark 1.3Similar price
  2. 1447
    GPT-5.4 MiniSimilar price
  3. 1443
    InklingSimilar price
  4. 1434
    GLM 5V TurboSimilar price
  5. 1415
    Mistral Large 3 2512Nearby rank
  6. 1415
    GPT-4.1Nearby rank
  7. 1414
    Claude Haiku 4.5
  8. 1413
    Qwen3 VL 235B A22B InstructNearby rank
  9. 1412
    Inkling 2 SmallNearby rank
  10. 1325
    Claude Haiku 3.5Predecessor

+89 vs Claude Haiku 3.5 (predecessor) · Ahead of 0 of 4 similarly priced models

Arena WebDev

Best Arena rating per model · Higher is better

Arena #132

  1. 1679
    GPT-6.1 TerraSimilar price
  2. 1655
    Muse Spark 1.3Similar price
  3. 1515
    Qwen3.7 MaxSimilar price
  4. 1399
    GLM 5V TurboSimilar price
  5. 1338
    GPT-5.1-CodexNearby rank
  6. 1331
    MiMo-V2-FlashNearby rank
  7. 1330
    Claude Haiku 4.5
  8. 1322
    Kimi K2 Thinking TurboNearby rank
  9. 1303
    Laguna XS.2Nearby rank

Ahead of 0 of 4 similarly priced models

Arena Vision

Best Arena rating per model · Higher is better

Not ranked on Arena Vision yet

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

Arena #44

  1. 1480
    Muse Spark 1.1Similar price
  2. 1469
    Gemini 3.8 FlashSimilar price
  3. 1450
    Kimi K2.6Similar price
  4. 1433
    Gemini 2.5 ProNearby rank
  5. 1425
    GPT-5.2Nearby rank
  6. 1419
    Claude Haiku 4.5
  7. 1416
    GPT-5.1Nearby rank
  8. 1406
    GLM 5V TurboSimilar price

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

#67 of 85

  1. 89.2%
    Gemini 3.8 FlashSimilar price
  2. 36.5%
    InklingSimilar price
  3. 22.8%
    GLM 5.2Similar price
  4. 18.9%
    GPT-5.4 MiniSimilar price
  5. 4.9%
    GLM 5Nearby rank
  6. 4.4%
    GPT-5 MiniNearby rank
  7. 4.0%
    Claude Haiku 4.5
  8. 4.0%
    DeepSeek V3.2Nearby rank
  9. 2.6%
    GPT-5 NanoNearby rank

Ahead of 0 of 4 similarly priced models

FrontierMath-2025-02-28-Private

Best score per model · Higher is better

#38 of 58

  1. 39.0%
    Kimi K2.6Similar price
  2. 33.4%
    GLM 5.1Similar price
  3. 28.3%
    GPT-5.4 MiniSimilar price
  4. 8.5%
    Qwen3 235B A22B Thinking 2507Similar price
  5. 7.2%
    Claude Opus 4.1Nearby rank
  6. 6.2%
    Qwen3.5-FlashNearby rank
  7. 5.9%
    Claude Haiku 4.5
  8. 5.5%
    GPT-4.1Nearby rank
  9. 4.8%
    Gemini 2.5 FlashNearby rank
  10. 0.3%
    Claude Haiku 3.5Predecessor

+5.6 pts vs Claude Haiku 3.5 (predecessor) · Ahead of 0 of 4 similarly priced models

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

Best score per model · Higher is better

#32 of 46

  1. 14.6%
    Kimi K2.6Similar price
  2. 12.5%
    GLM 5.1Similar price
  3. 2.1%
    GLM 5Nearby rank
  4. 2.1%
    Qwen3.5 PlusNearby rank
  5. 2.1%
    Claude Haiku 4.5
  6. 2.1%
    Gemini 2.5 Pro Preview 06-05Nearby rank
  7. 2.1%
    GPT-5 NanoNearby rank
  8. 2.1%
    GPT-5.4 MiniSimilar price
  9. 0.0%
    Qwen3 235B A22B Thinking 2507Similar price

Ahead of 1 of 4 similarly priced models

GPQA diamond

Best score per model · Higher is better

#105 of 161

  1. 94.3%
    GPT-6.1 TerraSimilar price
  2. 90.9%
    Qwen3.7 MaxSimilar price
  3. 89.9%
    GLM 5.1Similar price
  4. 88.3%
    InklingSimilar price
  5. 72.6%
    Qwen3 MaxNearby rank
  6. 71.5%
    Seed-OSS-36B-InstructNearby rank
  7. 71.2%
    Claude Haiku 4.5
  8. 71.2%
    DeepSeek V3Nearby rank
  9. 70.7%
    Qwen3 235B-A22BNearby rank
  10. 38.1%
    Claude Haiku 3.5Predecessor

+33.1 pts vs Claude Haiku 3.5 (predecessor) · Ahead of 0 of 4 similarly priced models

OTIS Mock AIME 2024-2025

Best score per model · Higher is better

#104 of 158

  1. 100.0%
    GPT-6.1 TerraSimilar price
  2. 99.2%
    Muse Spark 1.3Similar price
  3. 95.6%
    Qwen3.7 MaxSimilar price
  4. 93.3%
    GLM 5.1Similar price
  5. 67.5%
    Seed-OSS-36B-InstructNearby rank
  6. 66.9%
    Qwen3 32BNearby rank
  7. 66.7%
    Claude Haiku 4.5
  8. 66.4%
    R1 0528Nearby rank
  9. 66.4%
    Qwen3 14BNearby rank
  10. 4.3%
    Claude Haiku 3.5Predecessor

+62.4 pts vs Claude Haiku 3.5 (predecessor) · Ahead of 0 of 4 similarly priced models

SimpleQA Verified

Best score per model · Higher is better

#80 of 87

  1. 60.3%
    Muse Spark 1.2Similar price
  2. 57.3%
    GPT-6.1 TerraSimilar price
  3. 55.8%
    Qwen3.7 MaxSimilar price
  4. 34.0%
    GLM 5.1Similar price
  5. 17.4%
    Seed 1.6Nearby rank
  6. 15.9%
    Qwen3.6 FlashNearby rank
  7. 13.2%
    Claude Haiku 4.5
  8. 12.7%
    GPT-4.1 MiniNearby rank
  9. 11.7%
    GPT-5 NanoNearby rank

Ahead of 0 of 4 similarly priced models

Terminal Bench

Best score per model · Higher is better

#33 of 49

  1. 60.4%
    GPT-5.1-Codex-MaxSimilar price
  2. 57.3%
    Grok 4.20Similar price
  3. 52.4%
    GLM 5Similar price
  4. 36.6%
    MiniMax-M2.1Nearby rank
  5. 35.7%
    Kimi K2 ThinkingNearby rank
  6. 35.5%
    Claude Haiku 4.5
  7. 34.8%
    GPT-5 MiniNearby rank
  8. 33.4%
    GLM 4.7Nearby rank
  9. 32.6%
    Gemini 2.5 ProSimilar price

Ahead of 1 of 4 similarly priced models

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

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.