Inverse Scaling Prize Tasks
Models answer questions designed to reveal cases where larger models perform worse. The suite includes hindsight bias and following instructions that conflict with familiar text.
- Made by
- Ian R. McKenzie and colleagues
- Released
- June 2023
- Kind
- Language reasoning
- Level
- School
- Scored
- Checked automatically
- Answers
- Multiple choice
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Single turn
- Refresh
- One fixed set
- Run by
- Anyone
- Last model added
- April 2022
- Scores refreshed
- November 20th, 2026
PaLM (540B) leads with 78.2% PaLM (540B) leads with 78.2%
Best score per entry from Wei et al.’s leaderboard, read on November 20th, 2026. The score averages the ten accuracy tasks. Prompt injection uses loss and appears separately. 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.
- 1
PaLM (540B) 1-shot CoT 78.2%
- 2 PaLM 62B 1-shot CoT 68.2%
- 3 PaLM 8B 1-shot CoT 66%
- 4 PaLM 1B 1-shot (controlled) 53.1%
Source: Wei et al.’s leaderboard
- 1
PaLM (540B) 1-shot CoT 78.2%
- 2 PaLM 62B 1-shot CoT 68.2%
Source: Wei et al.’s leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 78.2% | 1-shot CoT | Apr 2022 | Official leaderboard |
| 2 | PaLM 62B | 68.2% | 1-shot CoT | Apr 2022 | Official leaderboard |
| 3 | PaLM 8B | 66% | 1-shot CoT | Official leaderboard | |
| 4 | PaLM 1B | 53.1% | 1-shot (controlled) | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 78.2% | 1-shot CoT | Apr 2022 | Official leaderboard |
| 2 | PaLM 62B | 68.2% | 1-shot CoT | Apr 2022 | Official leaderboard |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Wei et al.’s leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Wei et al.’s 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 Knowledge
- MMLU-Pro In use Harder 10-option version of MMLU with reasoning-focused questions across 14 subjects.
Intern-S2-397B 89.8%
- MMLU In use 57-subject multiple-choice exam questions from elementary to professional level. Scored by accuracy.
Seed1.8 92.3%
- SuperGPQA In use 26,529 graduate-level multiple-choice questions across 285 disciplines. Accuracy.
Qwen 3.6 Max (Preview) 73.9%
- WideSearch In use 200 bilingual broad info-seeking tasks: agent collects many atomic facts into a table. Scored by success rate and row/item F1.
Tencent Hy4 preview 83.9%
- BBEH micro average In use Models solve harder replacements for each of the 23 BIG-Bench Hard tasks, 4,520 examples in total. The score is the share of all examples answered correctly.
Gemma 4 31B IT 74.4%
- HLE-Diamond In use Models answer 1,000 cleaned expert questions from Humanity's Last Exam, 500 on reasoning and 500 on knowledge, closed book. Graded on exact answers. Claude Fable 5.2 77.9%
Data audit
8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 4 years ago, in April 2022, for PaLM (540B). 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 benchmark’s paper (4 entries).
- Note Models scored Only 2 models we track, plus 2 other entries on its leaderboard.
- Warning Newest score The newest came 4 years ago, in April 2022, for PaLM (540B).
- Note Frontier models None of its models are on the capabilities index.
- Pass Headroom The best is 78.2%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on Wei et al.: 10 rows for models we track and 2 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.