BenchLeader
Menu
Best LLMsChoose a modelCoding plansLeaderboardsModelsBenchmarksProvidersReleases7

LMArena Instruction Following leaderboard

As of 6 Nov 2026, the top score on LMArena Instruction Following is 1525, shared by Gemini 4 Carbon and Gemini 4 Argon, ahead of Claude Opus 5.5 at 1522, across 376 model configurations with a published result (Arena (LMArena), result dated 28 Oct 2026). Pairwise votes on prompts that set explicit constraints on the answer. LMArena Instruction Following is shown for reference only and does not count toward the BenchLeader Index.

Key facts (as of 6 Nov 2026)

  • Gemini 4 Carbon and Gemini 4 Argon share the top Arena Instruction Following score at 1525 (Arena (LMArena), 6 Nov 2026).
  • Arena Instruction Following: Text arena rating on prompts with explicit instructions (Arena (LMArena)).
  • 376 model configurations have a published Arena Instruction Following result on BenchLeader (BenchLeader, 6 Nov 2026).
  • Arena Instruction Following does not feed the BenchLeader Index; it is shown for reference (BenchLeader methodology).

Text arena rating on prompts with explicit instructions.

Published by
Arena (LMArena)
Category
Instruction following
Index weight
Reference only
Models
376
Data as of
6 Nov 2026

CC BY 4.0 — lmarena-ai/leaderboard-dataset on Hugging Face.

What the test looks like

Pairwise votes on prompts that set explicit constraints on the answer.

How it is scored

Bradley-Terry rating from pairwise votes, in Elo-like units, with style control.

What to keep in mind

Voters judge compliance by eye; see IFEval for programmatic checks.

Top 15
  1. 1Gemini 4 Carbon (high)1525
  2. 2Gemini 4 Argon (high)1525
  3. 3Claude Opus 5.5 (high)1522
  4. 4Claude Opus 4.6 (high)1511
  5. 5Claude Fable 5 (high)1509
  6. 6Claude Opus 4.7 (high)1502
  7. 7Claude Fable 5.1 (max)1500
  8. 8Muse Wonder (max)1498
  9. 9Claude Opus 4.61498
  10. 10Claude Opus 5 (high)1496
  11. 11Claude Opus 5 (max)1495
  12. 12Claude Opus 4.71493
  13. 13Claude Opus 4.8 (high)1492
  14. 14Gemini 3.8 Flash (high)1491
  15. 15GPT-5.6 Sol (xhigh)1491
376 of 376
#
1Gemini 4 Carbonhigh152569.22026-10-28
2Gemini 4 Argonhigh152566.02026-10-28
3Claude Opus 5.5high152265.82026-10-28
4Claude Opus 4.6high151162.62026-10-28
5Claude Fable 5high150964.32026-10-28
6Claude Opus 4.7high150260.92026-10-28
7Claude Fable 5.1max150067.22026-10-28
8Muse Wondermax149862.32026-10-28
9Claude Opus 4.6149862.32026-10-28
10Claude Opus 5high149664.32026-10-28
11Claude Opus 5max149565.42026-10-28
12Claude Opus 4.7149363.12026-10-28
13Claude Opus 4.8high149260.62026-10-28
14Gemini 3.8 Flashhigh149162.02026-10-28
15GPT-5.6 Solxhigh149164.22026-10-28
16Kimi K3max149062.12026-10-28
17Gemini 3.7 Flashhigh148762.22026-10-28
18Claude Sonnet 5.5xhigh148663.42026-10-28
19GPT-6.1 Solmax148667.32026-10-28
20Claude Opus 4.5high148459.02026-10-28
21Muse Spark 1.3max148362.02026-10-28
22Gemini 3.1 Pro148060.52026-10-28
23GLM 5.3max147959.02026-10-28
24Claude Opus 4.8147860.02026-10-28
25GPT-5.5high147863.72026-10-28
26DeepSeek V4.1 Flashmax147759.02026-10-28
27GPT-5.5147561.72026-10-28
28Qwen3.8 Maxmax147558.82026-10-28
29Muse Spark 1.2xhigh147561.72026-10-28
30Claude Sonnet 4.6147458.22026-10-28
31Claude Opus 4.5147457.92026-10-28
32GLM 5.3 Flash147456.82026-10-28
33Gemini 3.6 Flashhigh147361.22026-10-28
34Gemini 3 Pro147260.22026-10-28
35Muse Spark 1.1147262.22026-10-28
36GPT-6 Astramax147267.82026-10-28
37MiMo-V2.6-Pro147257.32026-10-28
38GPT-5.4high147158.82026-10-28
39GLM 5.2max146959.52026-10-28
40MiMo-V2.5-Pro146754.92026-10-28
41Qwen3.5 Maxmax1467–2026-10-28
42Claude Sonnet 5high146759.32026-10-28
43Gemini 3.5 Flashmedium146661.92026-10-28
44Gemini 3.5 Flashhigh146562.52026-10-28
45Qwen3.7max146560.32026-10-28
46Muse Spark146464.12026-10-28
47Claude Sonnet 4.5high146457.92026-10-28
48GPT-5.6 Terraxhigh146360.32026-10-28
49Claude Sonnet 4.5146253.72026-10-28
50DeepSeek V4 Prohigh146257.52026-10-28
51GLM 5.1146255.72026-10-28
52Grok 4.5146159.82026-10-28
53GPT-5.4146158.02026-10-28
54Claude Opus 4.1thinking145857.82026-10-28
55Gemini 3 Flash145856.92026-10-28
56GPT-5.2145757.92026-10-28
57Kimi K2.6145655.22026-10-28
58GPT-5.5 Instant145557.02026-10-28
59Claude Opus 4.1145451.42026-10-28
60Gemma 4 31B145447.82026-10-28

More instruction following benchmarks · All LLM benchmarks

Related benchmarks

Top head-to-heads on Arena Instruction Following

Cite as: BenchLeader, “Arena Instruction Following leaderboard”, data as of 6 Nov 2026, https://www.benchleader.com/benchmarks/lmarena_text_if.

Arena Instruction Following: questions

What does Arena Instruction Following measure?
Pairwise votes on prompts that set explicit constraints on the answer. Scores are reported in Elo-style ratings from pairwise comparisons; higher is better.
What is the top score on LMArena Instruction Following?
As of 6 Nov 2026, the highest score on Arena Instruction Following is 1525, shared by Gemini 4 Carbon and Gemini 4 Argon, ahead of Claude Opus 5.5 at 1522. Scores come from Arena (LMArena) and are re-read daily.
Which AI model leads Arena Instruction Following?
Gemini 4 Carbon and Gemini 4 Argon tie for first on Arena Instruction Following at 1525 as of 6 Nov 2026, ahead of Claude Opus 5.5 at 1522.
How many models have Arena Instruction Following results?
376 model configurations have a Arena Instruction Following result on BenchLeader, all taken from Arena (LMArena).
Who runs Arena Instruction Following and how often is it updated?
Arena Instruction Following is published by Arena (LMArena). BenchLeader re-reads the published results once a day and records the date each result was published.
Does Arena Instruction Following count toward the BenchLeader Index?
No. Arena Instruction Following is shown for reference but left out of the composite index.