Models

Models on the leaderboard

We test AI models that read a short text and make a quick judgment about a person. Does the model’s answer change when we change a name, a pronoun or another personal detail? Here you can compare the 3 models tested so far and open each one’s results.

Models are listed in a fixed order, not ranked: compare them characteristic by characteristic on each model's page and on the leaderboard. How the comparison works.

Jev
a clear effect on 9 of the 9 characteristics it was tested on · tested on 9 of 9
largest effect: 13.91 percentage points beyond the control edit
Jev is a hosted a fast decision model. We reach it through its maker's software kit, typesafe-sdk, on paid API access. We pay for that access and have no other relationship with its maker. The version tested is jev-1.13.0.
Laya
a clear effect on 9 of the 9 characteristics it was tested on · tested on 9 of 9
largest effect: 15.82 percentage points beyond the control edit
Laya is an open-source a fast decision model. We reached it at github.com/NandhaKishorM/laya (Apache-2.0). We ran two builds on our own Apple-silicon hardware. The independent Apple MLX port (laya-mlx 0.1.0) was used wherever we have it; the original PyTorch build (laya 0.3.7) was used where the MLX build has not run yet. The headline results matched, though probabilities can differ by up to about 0.02 on a single text.
Kev
a clear effect on 9 of the 9 characteristics it was tested on · tested on 9 of 9
largest effect: 15.15 percentage points beyond the control edit
Kev is an open-source a fast decision model. We reached it at github.com/jaredpalmer/kev. We ran it on our own Apple-silicon hardware with checkpoint jaredpalmer/kev-0.8b@54f4f8777356cd5bbbb6c6919c657f26e6f2f6d8, base model Qwen/Qwen3.5-0.8B-Base@dc7cdfe2ee4154fa7e30f5b51ca41bfa40174e68, MLX backend, bfloat16 precision.

Six things to know before you read the results

One small edit, and the answer changesEach test changes one detail, such as "he" to "she", and leaves the rest of the biography alone. Laya changed its answer on 17.85% of paralegal-or-attorney biographies when only that word changed. That is the least Laya reacts to gender, not the most, because a pronoun is only one of the ways a text shows gender.
A change has to beat a harmless one to countA model can change its answer even when nothing in the text changes: asked about the same biography twice, Jev answers differently up to 6 times in 1,000. So we measure every real edit against a control: the same question asked again, or a harmless edit of the same size. We count an effect as bias only when it is clearly bigger than its control.
Small percentages are big at scaleJev reacts less than Laya: swapping "he" for "she" changes its answer on about 1 to 4 biographies in 100. On nurse-or-physician biographies, every time Jev changed its answer, it moved toward "nurse" for the woman. Across a million screening decisions, 3.3% is 33,000 people.
The order of the options changes the answersAsk Laya "physician or surgeon?" instead of "surgeon or physician?" about the same biography, and Laya changes its answer on 6.6% of biographies. Every other number here keeps one fixed order for each question. The Option order results show how much the order matters.
Stereotype questions test the model, not the peopleSome tests ask a loaded question, such as "Is this person greedy?", about biographies that mention a religion or a nationality. A high stereotype score shows that the model has learned a stereotype. It says nothing about the people in the biographies.
Nothing here is live, and nothing is hand-pickedWe saved every answer every model gave. Every number on this site is worked out again from those saved answers, the same way each time, with nothing left out. You can check any figure in the data file.

Release v1.49.1, published 2026-09-30.

Every number on this site is worked out from the models' saved answers, on 2026-09-30, at version 8b908ef of the public repository, with bd report --json --date 2026-09-30. You can download every number in one data file. About this site.

Where the answers come from: studies/<task>-<cue>.jsonl (re-run from the saved answers with bd replay).

Biased-Decisions is MIT-licensed (source). Bias in Bios (De-Arteaga et al., 2019) is MIT-licensed on the Hugging Face Hub. Typefaces: Jersey 25 and Montserrat, SIL Open Font License.

Models on the leaderboard

Products

Open source we use

Research