Skip to content
The Aggregate

Unified LLM rankings, daily

Models Task Explorer Subscriptions What's New Trends Benchmarks Builder Guesswork
How It Works Daily brief
The Aggregate
Loading data...

BLINDSPOT - Over-Refusal Rate: leaderboard

Metric: ORR (%). Source: arxiv.org. 13 models tracked.

Top models

#ModelScore
1Mistral Large 333
2Claude Haiku 4.530
3GPT-3.5 Turbo27
4Gemma 3 12B23
5Gemma 3 27B18
6Claude Opus 4.617
7GPT-4o16
8GPT-4.513
9GPT-5.6 Terra12
10Gemini 2.5 Pro11
11Llama 3.3 70B10
12GPT-5.6 Luna9
13GPT-5.6 Sol7

Interactive version: theaggregate.ai/benchmark?slug=blindspot-over-refusal-rate · How It Works · Data refreshed daily, snapshot 2026-09-20.

Built by Mikhail Doroshenko — AI researcher, co-author of Humanity’s Last Exam. Independent project; no affiliation with any AI lab.
A

Your Aggregate

Sign in to keep your work

Save custom rankings and get 50 free task analyses each day.

Continue with Google Continue with GitHub

No password to remember. Your public profile identifies saved work.

Your Aggregate

Account

  • Saved custom aggregate views
  • 50 free task analyses per day
Sign out