Benchmark Almanac

Right Answer, Wrong Reason: ML Judgment Calls

Claude Sonnet 5.5 32.0 leads since October 2026
Not in model cards

Models make 36 decisions about tabular machine learning. The benchmark compares choosing from options with giving advice freely against measured outcomes.

Kaggle publishes scores for 10 distinct model versions.

Made by
starkhushi
Released
October 2026
Kind
Machine learning engineering
Level
University
Scored
Graded by a model
Answers
Open-ended
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
Strongest model tested
GPT-6.1 Sol #6 on the capability index
Last model added
September 2026
Scores refreshed
November 20th, 2026
10
Models scored on its own leaderboard

Claude Sonnet 5.5 leads with 32.0 among models released by September 2026

Best score per model from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best.

The newest model with a score here came out in September 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 Claude Sonnet 5.5 32.0
  2. 2 Grok 4.20 reasoning 31.0
  3. 3 Gemma 4 26B A4B 31.0
  4. 4 Gemini 3.8 Flash 31.0
  5. 5 Gemini 3.1 Pro 30.5
  6. 6 Gemma 4 31B IT 30.5
  7. 7 Gemini 3.5 Flash-Lite 29.0
  8. 8 GPT-6.1 Sol 29.0
  9. 9 DeepSeek-R1 May 2025 28.5
  10. 10 Claude Haiku 4.5 28.0

Source: Official Right Answer, Wrong Reason: ML Judgment Calls leaderboard

#ModelScoreSettingReleasedRun onSource
1 Claude Sonnet 5.5 32.0 Sep 2026 Oct 2026 Official leaderboard
2 Grok 4.20 31.0 reasoning Feb 2026 Oct 2026 Official leaderboard
3 Gemma 4 26B A4B 31.0 Apr 2026 Oct 2026 Official leaderboard
4 Gemini 3.8 Flash 31.0 Sep 2026 Oct 2026 Official leaderboard
5 Gemini 3.1 Pro 30.5 Feb 2026 Oct 2026 Official leaderboard
6 Gemma 4 31B IT 30.5 Apr 2026 Oct 2026 Official leaderboard
7 Gemini 3.5 Flash-Lite 29.0 Jul 2026 Oct 2026 Official leaderboard
8 GPT-6.1 Sol 29.0 Sep 2026 Oct 2026 Official leaderboard
9 DeepSeek-R1 May 2025 28.5 May 2025 Oct 2026 Official leaderboard
10 Claude Haiku 4.5 28.0 Oct 2025 Oct 2026 Official leaderboard

Number one over time

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

Anthropic
Own leaderboard Not tested since Oct 2026

Source: Official Right Answer, Wrong Reason: ML Judgment Calls 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
28.0 29.0 30.0 31.0 32.0 2025 2026 28.0 29.0 30.0 31.0 32.0 2025 2026

Source: Official Right Answer, Wrong Reason: ML Judgment Calls 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 AI research

All 164 in AI research
  • PaperBench In use Agent replicates ICML 2024 papers from scratch. LLM judge grades against author-written hierarchical rubrics. Qwen 3.8 Max 93%
  • MLS-Bench In use Agent must invent one algorithmic change in an ML research scaffold whose gain transfers across settings. Scored relative to human baselines. Claude Fable 5.1 50.3%
  • PostTrainBench 1.2 In use CLI agent gets a base model, one H100 and a time budget to post-train it for a target benchmark. Scored on normalized gain. Claude Fable 5.1 44.6%
  • MLE-bench In use Kaggle-style ML engineering competitions. Agent trains models and is scored against competition leaderboards. GPT-6.2 Astra 96.1%
  • PMPP-Hard In use Agents write GPU kernels for 69 tasks that must be correct, and on 48 of them also run within 1.25 times the reference time. The score is the area under the solved-tasks curve over a 10-hour budget, out of 690 task-hours. Kimi K3 0.71
  • KernelBench In use Replace PyTorch operators with faster custom GPU kernels. Scored on correctness and speedup. MiniMax-M3 28.8%

Data audit

All 9 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (10 entries).
  • Pass Models scored 10 models we track.
  • Pass Newest score The newest is from September 2026, for GPT-6.1 Sol.
  • Note Frontier models Only 2 of the top 10 have a score. The strongest tested is GPT-6.1 Sol (#6).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on kaggle.com: 10 rows for models we track 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