# Elorian — Introducing DecoyCount

- Company: Elorian (elorian.ai)
- Announced: 2026-10-29
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://elorian.ai/blog/decoycount
- Record: https://forck.live/items/30691-introducing-decoycount
- Subject: Glimpse visual reasoning models
- Models affected: Claude Opus 5, GPT 5.6 Sol, Gemini 3.6 Flash, GPT 6 Astra, Gemini 3.8 Flash

Elorian introduced DecoyCount, a benchmark of 1,660 real-world images built to separate perception from counting by recording the specific wrong count each scene invites. Each image carries a pre-registered decoy, a wrong answer the scene suggests, logged in advance with the mechanism that invites it, so a model that returns the decoy reveals which interpretation it chose. The benchmark spans 2,203 decoys across seven mechanisms, with three prompts per image: a neutral question, a cued version naming the intended rule, and a multi-hop scenario that implies the rule without stating it. Items were kept only if Claude Opus 5, GPT 5.6 Sol and Gemini 3.6 Flash all missed the neutral answer and at least one returned a registered decoy exactly; blind review retained 1,660 of 2,100 candidates, corrected 121 counts and replaced 79 images. Elorian reports that a one-sentence cue moves accuracy by 11.8 to 22.1 points, and that changing "are there" to "are visible" shifts Gemini 3.8 Flash by 6.69 points, repairing 246 answers while breaking 135. The best unaided score was 43.2% exact match, by GPT 6 Astra, and models answered the decoy five of six times more often than the truth.

## Evidence

Verbatim from https://elorian.ai/blog/decoycount:

> Decoys 2,203 pre-registered, 7 mechanisms
> 
> Prompts 3 per image, same count
> 
> Best unaided 43.2% GPT 6 Astra, exact match
> 
> Answer the decoy 5 of 6 more often than the truth
> 
> Counting is a fundamental step in complex problem solving. Models today can’t count reliably.
> 
> The task
> 
> How many bottles are there?
> 
> What answering it actually takes
> 
> Interpret the prompt
> 
> Review the scene and locate the target
> 
> Decide what counts as one instance, and in which unit
> 
> Tell physical objects apart from any similar objects
> 
> Identify depictions and printed text
> 
> Enumerate what is left
> 
> Each of those steps can fail on its own, and a single final number answer collapses all of the step grading into one score. So a wrong count tells you almost nothing on its own. This case is printed 12 and holds more than twelve bottles. Answer of twelve shows the model read a number and trusted it over the scene.
> 
> DecoyCount is uniquely built to separate perception from the counting task. It is 1,660 real-world images, each carrying a pre-registered decoy . The decoy is a specific wrong count that the scene invites, recorded in advance along with the mechanism that invites it.

---

Record: https://forck.live/items/30691-introducing-decoycount
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md