OpenGPTDécisions fondées sur des preuves
Commencer l’évaluation

July 2026 · Research edition

Classement des modèles d’IA pour des décisions réelles

See which exact model configurations lead on independent intelligence, objective-task, cost, and open-weight evidence—then compare two on your own work.

Compare two leading modelsRead the scoring method
Published
24 juil. 2026
Data cutoff
24 juil. 2026
Method
OpenGPT research synthesis 1.0
Base du classement
3 independent public sources

July 2026 · Research edition

Choisissez un angle de classement

378 configurations publiées

Préférence humaine

LMArena’s latest style-controlled overall ranking. Every published configuration in the source snapshot is included; use search and filters to explore the full list.

Independent sourceLMArena
Source release: Text Style Control · latest · 2026-07-21Captured: 24 juil. 2026Open source results ↗Read source methodology ↗

40 sur 378 affichées. Préférence humaine.

  1. Classement#1

    claude-fable-5

    Anthropic
    Source configuration label
    Text · Style control · Overall
    OpenGPT catalog ID
    claude-fable-5
    Votes
    14 646
    Confidence interval
    1500.9–1513.7
    License
    Proprietary
    Préférence humaine1507
    Accéder au modèle
  2. Classement#2

    claude-opus-4-6-thinking

    Anthropic
    Source configuration label
    Text · Style control · Overall
    Votes
    63 191
    Confidence interval
    1501.1–1508.6
    License
    Proprietary
    Préférence humaine1505
    Not yet in OpenGPT’s exact-model catalog
  3. Classement#3

    claude-opus-4-7-thinking

    Anthropic
    Source configuration label
    Text · Style control · Overall
    Votes
    50 683
    Confidence interval
    1497.8–1506.2
    License
    Proprietary
    Préférence humaine1502
    Not yet in OpenGPT’s exact-model catalog
  4. Classement#4

    claude-opus-4-6

    Anthropic
    Source configuration label
    Text · Style control · Overall
    Votes
    67 037
    Confidence interval
    1494.1–1501.4
    License
    Proprietary
    Préférence humaine1498
    Not yet in OpenGPT’s exact-model catalog
  5. Classement#5

    muse-spark-1.1

    Meta
    Source configuration label
    Text · Style control · Overall
    Votes
    7 927
    Confidence interval
    1487.8–1502.3
    License
    Proprietary
    Préférence humaine1495
    Not yet in OpenGPT’s exact-model catalog
  6. Classement#6

    claude-opus-4-7

    Anthropic
    Source configuration label
    Text · Style control · Overall
    Votes
    51 788
    Confidence interval
    1489.5–1497.9
    License
    Proprietary
    Préférence humaine1494
    Not yet in OpenGPT’s exact-model catalog
  7. Classement#7

    muse-spark

    Meta
    Source configuration label
    Text · Style control · Overall
    Votes
    13 565
    Confidence interval
    1481.7–1493.4
    License
    Proprietary
    Préférence humaine1488
    Not yet in OpenGPT’s exact-model catalog
  8. Classement#8

    gemini-3.1-pro-preview

    Google
    Source configuration label
    Text · Style control · Overall
    OpenGPT catalog ID
    gemini-3.1-pro-preview
    Votes
    84 631
    Confidence interval
    1482.3–1489.3
    License
    Proprietary
    Préférence humaine1486
    Accéder au modèle
  9. Classement#9

    gemini-3-pro

    Google
    Source configuration label
    Text · Style control · Overall
    Votes
    41 268
    Confidence interval
    1481.9–1489.6
    License
    Proprietary
    Préférence humaine1486
    Not yet in OpenGPT’s exact-model catalog
  10. Classement#10

    kimi-k3

    Moonshot AI
    Source configuration label
    Text · Style control · Overall
    OpenGPT catalog ID
    kimi-k3
    Votes
    3 619
    Confidence interval
    1475.7–1495.6
    License
    Proprietary
    Préférence humaine1486
    Accéder au modèle
  11. Classement#11

    gpt-5.6-sol-xhigh

    OpenAI
    Source configuration label
    Text · Style control · Overall
    Votes
    6 221
    Confidence interval
    1477.2–1493.0
    License
    Proprietary
    Préférence humaine1485
    Not yet in OpenGPT’s exact-model catalog
  12. Classement#12

    gemini-3.6-flash

    Google
    Source configuration label
    Text · Style control · Overall
    Votes
    4 747
    Confidence interval
    1476.2–1493.9
    License
    Proprietary
    Préférence humaine1485
    Not yet in OpenGPT’s exact-model catalog
  13. Classement#13

    claude-opus-4-8-thinking

    Anthropic
    Source configuration label
    Text · Style control · Overall
    Votes
    30 901
    Confidence interval
    1478.5–1488.7
    License
    Proprietary
    Préférence humaine1484
    Not yet in OpenGPT’s exact-model catalog
  14. Classement#14

    gpt-5.5-high

    OpenAI
    Source configuration label
    Text · Style control · Overall
    Votes
    45 760
    Confidence interval
    1477.3–1486.1
    License
    Proprietary
    Préférence humaine1482
    Not yet in OpenGPT’s exact-model catalog
  15. Classement#15

    gpt-5.4-high

    OpenAI
    Source configuration label
    Text · Style control · Overall
    Votes
    58 997
    Confidence interval
    1473.8–1481.7
    License
    Proprietary
    Préférence humaine1478
    Not yet in OpenGPT’s exact-model catalog
  16. Classement#16

    gpt-5.5

    OpenAI
    Source configuration label
    Text · Style control · Overall
    Votes
    47 180
    Confidence interval
    1472.0–1480.7
    License
    Proprietary
    Préférence humaine1476
    Not yet in OpenGPT’s exact-model catalog
  17. Classement#17

    gemini-3.5-flash-high

    Google
    Source configuration label
    Text · Style control · Overall
    Votes
    10 092
    Confidence interval
    1469.7–1482.8
    License
    Proprietary
    Préférence humaine1476
    Not yet in OpenGPT’s exact-model catalog
  18. Classement#18

    gpt-5.2-chat-latest-20260210

    OpenAI
    Source configuration label
    Text · Style control · Overall
    Votes
    34 420
    Confidence interval
    1471.6–1479.8
    License
    Proprietary
    Préférence humaine1476
    Not yet in OpenGPT’s exact-model catalog
  19. Classement#19

    qwen3.7-max-preview

    Alibaba
    Source configuration label
    Text · Style control · Overall
    Votes
    3 714
    Confidence interval
    1465.1–1485.2
    License
    Proprietary
    Préférence humaine1475
    Not yet in OpenGPT’s exact-model catalog
  20. Classement#20

    grok-4.20-beta1

    xAI
    Source configuration label
    Text · Style control · Overall
    Votes
    26 822
    Confidence interval
    1469.7–1479.0
    License
    Proprietary
    Préférence humaine1474
    Not yet in OpenGPT’s exact-model catalog
  21. Classement#21

    gemini-3.5-flash-medium

    Google
    Source configuration label
    Text · Style control · Overall
    Votes
    14 063
    Confidence interval
    1468.2–1480.4
    License
    Proprietary
    Préférence humaine1474
    Not yet in OpenGPT’s exact-model catalog
  22. Classement#22

    claude-opus-4-8

    Anthropic
    Source configuration label
    Text · Style control · Overall
    OpenGPT catalog ID
    claude-opus-4-8
    Votes
    31 619
    Confidence interval
    1468.3–1478.5
    License
    Proprietary
    Préférence humaine1473
    Accéder au modèle
  23. Classement#23

    gpt-5.5-instant

    OpenAI
    Source configuration label
    Text · Style control · Overall
    Votes
    25 995
    Confidence interval
    1468.2–1478.5
    License
    Proprietary
    Préférence humaine1473
    Not yet in OpenGPT’s exact-model catalog
  24. Classement#24

    grok-4.20-beta-0309-reasoning

    xAI
    Source configuration label
    Text · Style control · Overall
    Votes
    60 330
    Confidence interval
    1469.3–1477.0
    License
    Proprietary
    Préférence humaine1473
    Not yet in OpenGPT’s exact-model catalog
  25. Classement#25

    gemini-3-flash

    Google
    Source configuration label
    Text · Style control · Overall
    Votes
    30 682
    Confidence interval
    1468.7–1477.5
    License
    Proprietary
    Préférence humaine1473
    Not yet in OpenGPT’s exact-model catalog
  26. Classement#26

    claude-opus-4-5-20251101-thinking-32k

    Anthropic
    Source configuration label
    Text · Style control · Overall
    Votes
    37 037
    Confidence interval
    1469.0–1476.8
    License
    Proprietary
    Préférence humaine1473
    Not yet in OpenGPT’s exact-model catalog
  27. Classement#27

    claude-sonnet-4-6

    Anthropic
    Source configuration label
    Text · Style control · Overall
    Votes
    57 183
    Confidence interval
    1468.7–1476.4
    License
    Proprietary
    Préférence humaine1473
    Not yet in OpenGPT’s exact-model catalog
  28. Classement#28

    grok-4.20-multi-agent-beta-0309

    xAI
    Source configuration label
    Text · Style control · Overall
    Votes
    59 116
    Confidence interval
    1467.1–1474.8
    License
    Proprietary
    Préférence humaine1471
    Not yet in OpenGPT’s exact-model catalog
  29. Classement#29

    glm-5.1

    Z.AI
    Source configuration label
    Text · Style control · Overall
    Votes
    30 726
    Confidence interval
    1465.1–1474.2
    License
    MIT
    Préférence humaine1470
    Not yet in OpenGPT’s exact-model catalog
  30. Classement#30

    glm-5.2 (max)

    Z.AI
    Source configuration label
    Text · Style control · Overall
    Votes
    18 017
    Confidence interval
    1463.4–1475.0
    License
    MIT
    Préférence humaine1469
    Not yet in OpenGPT’s exact-model catalog
  31. Classement#31

    claude-opus-4-5-20251101

    Anthropic
    Source configuration label
    Text · Style control · Overall
    Votes
    70 976
    Confidence interval
    1465.9–1472.3
    License
    Proprietary
    Préférence humaine1469
    Not yet in OpenGPT’s exact-model catalog
  32. Classement#32

    ernie-5.1

    Baidu
    Source configuration label
    Text · Style control · Overall
    Votes
    37 418
    Confidence interval
    1463.1–1472.4
    License
    Proprietary
    Préférence humaine1468
    Not yet in OpenGPT’s exact-model catalog
  33. Classement#33

    grok-4.5

    xAI
    Source configuration label
    Text · Style control · Overall
    OpenGPT catalog ID
    grok-4.5
    Votes
    7 833
    Confidence interval
    1460.4–1475.0
    License
    Proprietary
    Préférence humaine1468
    Accéder au modèle
  34. Classement#34

    mimo-v2.5-pro

    Xiaomi
    Source configuration label
    Text · Style control · Overall
    Votes
    42 195
    Confidence interval
    1462.4–1471.3
    License
    MIT
    Préférence humaine1467
    Not yet in OpenGPT’s exact-model catalog
  35. Classement#35

    gpt-5.4

    OpenAI
    Source configuration label
    Text · Style control · Overall
    Votes
    62 036
    Confidence interval
    1462.4–1470.1
    License
    Proprietary
    Préférence humaine1466
    Not yet in OpenGPT’s exact-model catalog
  36. Classement#36

    grok-4.1-thinking

    xAI
    Source configuration label
    Text · Style control · Overall
    Votes
    65 461
    Confidence interval
    1462.7–1469.1
    License
    Proprietary
    Préférence humaine1466
    Not yet in OpenGPT’s exact-model catalog
  37. Classement#37

    qwen3.5-max-preview

    Alibaba
    Source configuration label
    Text · Style control · Overall
    Votes
    21 479
    Confidence interval
    1459.8–1469.8
    License
    Proprietary
    Préférence humaine1465
    Not yet in OpenGPT’s exact-model catalog
  38. Classement#38

    claude-sonnet-5-high

    Anthropic
    Source configuration label
    Text · Style control · Overall
    Votes
    13 521
    Confidence interval
    1454.9–1467.4
    License
    Proprietary
    Préférence humaine1461
    Not yet in OpenGPT’s exact-model catalog
  39. Classement#39

    kimi-k2.6

    Moonshot AI
    Source configuration label
    Text · Style control · Overall
    OpenGPT catalog ID
    kimi-k2.6
    Votes
    37 686
    Confidence interval
    1456.4–1465.5
    License
    Modified MIT
    Préférence humaine1461
    Accéder au modèle
  40. Classement#40

    qwen3.6-max-preview

    Alibaba
    Source configuration label
    Text · Style control · Overall
    Votes
    5 188
    Confidence interval
    1451.7–1468.5
    License
    Proprietary
    Préférence humaine1460
    Not yet in OpenGPT’s exact-model catalog
40 sur 378 affichées

Base du classement

How to read this research edition

  • Scores stay on their original publisher scales. They are not combined into a composite score.

  • Missing data does not mean zero. It means the publisher did not report that metric for this snapshot.

  • A position applies to the named model configuration and source release—not every task, region, or deployment.

  • No provider can pay for placement. OpenGPT links directly to the evidence used.

Base du classement

Data and sources

Each view preserves the source’s metric, release, and methodology. Follow the links to audit the underlying work.
AR

LMArena

The complete latest style-controlled overall ranking: all 378 published model configurations, based on human preference votes.

Source release
Text Style Control · latest · 2026-07-21
Captured
24 juil. 2026
License
CC BY 4.0

Open source results ↗Read source methodology ↗

Method

A predictable publishing rhythm

OpenGPT will separate changing signals from frozen editions so past decisions remain auditable.
  1. 01

    Weekly watch

    Review new model and benchmark releases; flag material changes without silently rewriting this edition.

  2. 02

    Monthly frozen edition

    Publish a dated snapshot with exact configurations, source releases, and corrections.

  3. 03

    Quarterly method review

    Review categories, source quality, conflicts, and inclusion rules before changing the method.

Published

Past snapshots

August 2026August 2026

Scheduled next edition. It will be published only after source capture and review.