Benchmark Almanac

Judge Arena (Atla)

Llama 3.3 70B Instruct 1,335 leads since December 2024
Not in model cards

Users compare two anonymous model judges that score and critique the same answer.

Pairwise votes determine the Elo ranking of the judges.

Made by
Atla
Released
November 2024
Kind
Other
Level
University
Scored
Ranked by votes
Answers
Open-ended
Tasks
All public
Measures
Preference
In and out
Text → Text
Interaction
Single turn
Items from
Live play or votes
Refresh
Live
Run by
A host
In views
Kept fresh
Strongest model tested
Claude 3.5 Sonnet #158 on the capability index
Last model added
March 2025
Scores refreshed
November 20th, 2026
28
Entries on its own leaderboard, 17 models we track

Llama 3.3 70B Instruct leads with 1,335 among models released by March 2025 Llama 3.3 70B Instruct leads with 1,335 among models released by March 2025

Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.

The newest model with a score here came out in March 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

Bars start at 900.

  1. 1 Llama 3.3 70B Instruct 1,335
  2. 2 Flow-Judge-v0.1 1,335
  3. 3 Claude 3 Opus 1,312
  4. 4 GPT-4o Mar 2025 1,308
  5. 5 Atla-8B-preview 1,306
  6. 6 GPT-4 Turbo Apr 2024 1,304
  7. 7 Claude 3 Haiku 1,286
  8. 8 Claude 3.5 Haiku 1,282
  9. 9 Qwen 2.5 7B Instruct Turbo 1,280
  10. 10 GPT-3.5 Turbo Jan 2024 1,271
  11. 11 Qwen2.5-72B 1,269
  12. 12 Meta Llama 3.1 405B Instruct Turbo 1,263
  13. 13 Meta Llama 3.1 8B Instruct Turbo 1,233
  14. 14 Atla Selene 1 1,216
  15. 15 Mistral (7B) Instruct v0.3 1,215
  16. 16 Claude 3.5 Sonnet 1,211
  17. 17 Atla Selene 1 Mini 1,204
  18. 18 QwQ-32B-Preview 1,172
  19. 19 Mistral-7B-Instruct-v0.1 1,157
  20. 20 Prometheus-7b v2 1,145
  21. 21 Qwen2-72B 1,123
  22. 22 Command R+ 1,119
  23. 23 Meta Llama 3.1 70B Instruct Turbo 1,111
  24. 24 Claude 3 Sonnet 1,049
  25. 25 Command R (03-2024) 1,040
  26. 26 Gemma 2 9B 1,035
  27. 27 Gemma 2 27B 1,013
  28. 28 SFR-LLaMA-3.1-70B-Judge 991

Source: Official Judge Arena (Atla) leaderboard

#EntryScore ±SettingReleasedRun onSource
1 Llama 3.3 70B Instruct 1,335 ±71 Dec 2024 Jan 2026 Official leaderboard
2 Flow-Judge-v0.1 open weights 1,335 ±102 Jan 2026 Official leaderboard
3 Claude 3 Opus 1,312 ±37 Feb 2024 Jan 2026 Official leaderboard
4 GPT-4o Mar 2025 1,308 ±37 Mar 2025 Jan 2026 Official leaderboard
5 Atla-8B-preview open weights 1,306 ±48 Jan 2026 Official leaderboard
6 GPT-4 Turbo Apr 2024 1,304 ±36 Apr 2024 Jan 2026 Official leaderboard
7 Claude 3 Haiku 1,286 ±37 Mar 2024 Jan 2026 Official leaderboard
8 Claude 3.5 Haiku 1,282 ±37 Oct 2024 Jan 2026 Official leaderboard
9 Qwen 2.5 7B Instruct Turbo open weights 1,280 ±36 Jan 2026 Official leaderboard
10 GPT-3.5 Turbo Jan 2024 1,271 ±36 Jan 2024 Jan 2026 Official leaderboard
11 Qwen2.5-72B 1,269 ±37 Sep 2024 Jan 2026 Official leaderboard
12 Meta Llama 3.1 405B Instruct Turbo open weights 1,263 ±35 Jan 2026 Official leaderboard
13 Meta Llama 3.1 8B Instruct Turbo open weights 1,233 ±36 Jan 2026 Official leaderboard
14 Atla Selene 1 closed weights 1,216 ±132 Jan 2026 Official leaderboard
15 Mistral (7B) Instruct v0.3 open weights 1,215 ±37 Jan 2026 Official leaderboard
16 Claude 3.5 Sonnet 1,211 ±38 Jun 2024 Jan 2026 Official leaderboard
17 Atla Selene 1 Mini open weights 1,204 ±217 Jan 2026 Official leaderboard
18 QwQ-32B-Preview 1,172 ±70 Nov 2024 Jan 2026 Official leaderboard
19 Mistral-7B-Instruct-v0.1 1,157 ±37 Sep 2023 Jan 2026 Official leaderboard
20 Prometheus-7b v2 open weights 1,145 ±84 Jan 2026 Official leaderboard
21 Qwen2-72B 1,123 ±37 Jun 2024 Jan 2026 Official leaderboard
22 Command R+ 1,119 ±53 Aug 2024 Jan 2026 Official leaderboard
23 Meta Llama 3.1 70B Instruct Turbo open weights 1,111 ±39 Jan 2026 Official leaderboard
24 Claude 3 Sonnet 1,049 ±36 Feb 2024 Jan 2026 Official leaderboard
25 Command R (03-2024) 1,040 ±56 Mar 2024 Jan 2026 Official leaderboard
26 Gemma 2 9B 1,035 ±36 Jun 2024 Jan 2026 Official leaderboard
27 Gemma 2 27B 1,013 ±37 Jun 2024 Jan 2026 Official leaderboard
28 SFR-LLaMA-3.1-70B-Judge closed weights 991 ±57 Jan 2026 Official leaderboard
#ModelScore ±SettingReleasedRun onSource
1 Llama 3.3 70B Instruct 1,335 ±71 Dec 2024 Jan 2026 Official leaderboard
2 Claude 3 Opus 1,312 ±37 Feb 2024 Jan 2026 Official leaderboard
3 GPT-4o Mar 2025 1,308 ±37 Mar 2025 Jan 2026 Official leaderboard
4 GPT-4 Turbo Apr 2024 1,304 ±36 Apr 2024 Jan 2026 Official leaderboard
5 Claude 3 Haiku 1,286 ±37 Mar 2024 Jan 2026 Official leaderboard
6 Claude 3.5 Haiku 1,282 ±37 Oct 2024 Jan 2026 Official leaderboard
7 GPT-3.5 Turbo Jan 2024 1,271 ±36 Jan 2024 Jan 2026 Official leaderboard
8 Qwen2.5-72B 1,269 ±37 Sep 2024 Jan 2026 Official leaderboard
9 Claude 3.5 Sonnet 1,211 ±38 Jun 2024 Jan 2026 Official leaderboard
10 QwQ-32B-Preview 1,172 ±70 Nov 2024 Jan 2026 Official leaderboard
11 Mistral-7B-Instruct-v0.1 1,157 ±37 Sep 2023 Jan 2026 Official leaderboard
12 Qwen2-72B 1,123 ±37 Jun 2024 Jan 2026 Official leaderboard
13 Command R+ 1,119 ±53 Aug 2024 Jan 2026 Official leaderboard
14 Claude 3 Sonnet 1,049 ±36 Feb 2024 Jan 2026 Official leaderboard
15 Command R (03-2024) 1,040 ±56 Mar 2024 Jan 2026 Official leaderboard
16 Gemma 2 9B 1,035 ±36 Jun 2024 Jan 2026 Official leaderboard
17 Gemma 2 27B 1,013 ±37 Jun 2024 Jan 2026 Official leaderboard

2 models have held number one on the official leaderboard since its release in November 2024

Lab claims count from the day the card came out, independent runs from the model’s release.

Anthropic Meta
Own leaderboard Not tested since Mar 2025

Source: Official Judge Arena (Atla) leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard
1,000 1,100 1,200 1,300 1,400 2024 2025 1,000 1,100 1,200 1,300 1,400 2024 2025

Source: Official Judge Arena (Atla) leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Other

All 69 in Other
  • Artificial Analysis Intelligence Index v5 In use Version 5 of Artificial Analysis's index: nine evaluations in four equal categories, most with private test sets, run three times each in its own agent harness. Claude Fable 5.2 64.6
  • Personal Superintelligence Index (Meta) In use Meta's composite of seven personal and knowledge-work evals: Agents' Last Exam, AutomationBench-AA, FocusBench, JobBench, GDP.pdf, VBT and DRACO.
  • Vals Science Index In use Agents are scored on the equal average of BioMysteryBench, MysteryMechanism and Terminal-Bench-Science. GPT-6.2 Astra 68.4%
  • Decision Index Public v0.3 In use Decision models return probabilities over labeled options in public tasks covering knowledge, language, retrieval, tools, and arts. A chance-corrected weighted aggregate combines the task scores. Granite Switch 4.1 30B 51.3
  • SAGE v1.1 (Vals) In use Models grade handwritten undergraduate math and computer science work against rubrics. Scored by class-balanced accuracy of each rubric check against expert grading. Claude Opus 5.5 91.3%
  • MTEB Fading Massive Text Embedding Benchmark: retrieval, clustering, classification across languages and modalities. Qwen3-Embedding-8B 72.2%

Data audit

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 20 months ago, in March 2025, for GPT-4o. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (28 entries).
  • Pass Models scored 17 models we track, plus 11 other entries on its leaderboard.
  • Warning Newest score The newest came 20 months ago, in March 2025, for GPT-4o.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Claude 3.5 Sonnet (#158).
  • Note Headroom Scored as an Elo rating, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on huggingface.co: 17 rows for models we track and 11 other entries at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check