RiddleBench: A New Generative Reasoning Benchmark for LLMs

Deepon Halder; Alan Saji; Thanmay Jayakumar; Anoop Kunchukuttan; Ratish Puduppully; Raj Dabre

RiddleBench: A New Generative Reasoning Benchmark for LLMs

Deepon Halder, Alan Saji, Thanmay Jayakumar, Anoop Kunchukuttan, Ratish Puduppully, Raj Dabre

Correct Metadata for

Use this form to create a GitHub issue with structured data describing the correction. You will need a GitHub account. Once you create that issue, the correction will be reviewed by a staff member.

⚠️ Mobile Users: Submitting this form to create a new issue will only work with github.com, not the GitHub Mobile app.

Important: The Anthology treat PDFs as authoritative. Please use this form only to correct data that is out of line with the PDF. See our corrections guidelines if you need to change the PDF.

Title Adjust the title. Retain tags such as <fixed-case>.

Authors Adjust author names and order to match the PDF.

Abstract Correct abstract if needed. Retain XML formatting tags such as <tex-math>. You may use <b>...</b> for bold, <i>...</i> for italic, and <url>...</url> for URLs.

Verification against PDF Ensure that the new title/authors match the snapshot below. (If there is no snapshot or it is too small, consult the PDF.)

Authors concatenated from the text boxes above:

ALL author names match the snapshot above—including middle initials, hyphens, and accents.

Abstract

While Large Language Models (LLMs) show remarkable capabilities, their complex reasoning skills require deeper investigation. We introduce **RiddleBench**, a new benchmark of 1,737 challenging puzzles designed to test reasoning beyond simple pattern matching. Our evaluation of state-of-the-art models reveals significant limitations, including hallucination cascades (uncritically accepting flawed peer reasoning) and poor self-correction due to strong self-confirmation bias. We also find that model performance is fragile, degrading when faced with reordered constraints or irrelevant information. RiddleBench serves as a resource for diagnosing these issues and guiding the development of more robust LLMs.

Anthology ID:: 2026.findings-eacl.228
Volume:: Findings of the Association for Computational Linguistics: EACL 2026
Month:: March
Year:: 2026
Address:: Rabat, Morocco
Editors:: Vera Demberg, Kentaro Inui, Lluís Marquez
Venue:: Findings
SIG:
Publisher:: Association for Computational Linguistics
Note:
Pages:: 4363–4372
Language:
URL:: https://aclanthology.org/2026.findings-eacl.228/
DOI:
Bibkey:
Cite (ACL):: Deepon Halder, Alan Saji, Thanmay Jayakumar, Anoop Kunchukuttan, Ratish Puduppully, and Raj Dabre. 2026. RiddleBench: A New Generative Reasoning Benchmark for LLMs. In Findings of the Association for Computational Linguistics: EACL 2026, pages 4363–4372, Rabat, Morocco. Association for Computational Linguistics.
Cite (Informal):: RiddleBench: A New Generative Reasoning Benchmark for LLMs (Halder et al., Findings 2026)
Copy Citation:
PDF:: https://aclanthology.org/2026.findings-eacl.228.pdf
Checklist:: 2026.findings-eacl.228.checklist.pdf

PDF Cite Search Checklist Fix data

Export citation

BibTeX
MODS XML
Endnote
Preformatted

@inproceedings{halder-etal-2026-riddlebench,
    title = "{R}iddle{B}ench: A New Generative Reasoning Benchmark for {LLM}s",
    author = "Halder, Deepon  and
      Saji, Alan  and
      Jayakumar, Thanmay  and
      Kunchukuttan, Anoop  and
      Puduppully, Ratish  and
      Dabre, Raj",
    editor = "Demberg, Vera  and
      Inui, Kentaro  and
      Marquez, Llu{\'i}s",
    booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EACL} 2026",
    month = mar,
    year = "2026",
    address = "Rabat, Morocco",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.findings-eacl.228/",
    pages = "4363--4372",
    ISBN = "979-8-89176-386-9",
    abstract = "While Large Language Models (LLMs) show remarkable capabilities, their complex reasoning skills require deeper investigation. We introduce **RiddleBench**, a new benchmark of 1,737 challenging puzzles designed to test reasoning beyond simple pattern matching. Our evaluation of state-of-the-art models reveals significant limitations, including hallucination cascades (uncritically accepting flawed peer reasoning) and poor self-correction due to strong self-confirmation bias. We also find that model performance is fragile, degrading when faced with reordered constraints or irrelevant information. RiddleBench serves as a resource for diagnosing these issues and guiding the development of more robust LLMs."
}

Download as File

<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="halder-etal-2026-riddlebench">
    <titleInfo>
        <title>RiddleBench: A New Generative Reasoning Benchmark for LLMs</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Deepon</namePart>
        <namePart type="family">Halder</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Alan</namePart>
        <namePart type="family">Saji</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Thanmay</namePart>
        <namePart type="family">Jayakumar</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Anoop</namePart>
        <namePart type="family">Kunchukuttan</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Ratish</namePart>
        <namePart type="family">Puduppully</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Raj</namePart>
        <namePart type="family">Dabre</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2026-03</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Findings of the Association for Computational Linguistics: EACL 2026</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Vera</namePart>
            <namePart type="family">Demberg</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Kentaro</namePart>
            <namePart type="family">Inui</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Lluís</namePart>
            <namePart type="family">Marquez</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Rabat, Morocco</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-386-9</identifier>
    </relatedItem>
    <abstract>While Large Language Models (LLMs) show remarkable capabilities, their complex reasoning skills require deeper investigation. We introduce **RiddleBench**, a new benchmark of 1,737 challenging puzzles designed to test reasoning beyond simple pattern matching. Our evaluation of state-of-the-art models reveals significant limitations, including hallucination cascades (uncritically accepting flawed peer reasoning) and poor self-correction due to strong self-confirmation bias. We also find that model performance is fragile, degrading when faced with reordered constraints or irrelevant information. RiddleBench serves as a resource for diagnosing these issues and guiding the development of more robust LLMs.</abstract>
    <identifier type="citekey">halder-etal-2026-riddlebench</identifier>
    <location>
        <url>https://aclanthology.org/2026.findings-eacl.228/</url>
    </location>
    <part>
        <date>2026-03</date>
        <extent unit="page">
            <start>4363</start>
            <end>4372</end>
        </extent>
    </part>
</mods>
</modsCollection>

Download as File

%0 Conference Proceedings
%T RiddleBench: A New Generative Reasoning Benchmark for LLMs
%A Halder, Deepon
%A Saji, Alan
%A Jayakumar, Thanmay
%A Kunchukuttan, Anoop
%A Puduppully, Ratish
%A Dabre, Raj
%Y Demberg, Vera
%Y Inui, Kentaro
%Y Marquez, Lluís
%S Findings of the Association for Computational Linguistics: EACL 2026
%D 2026
%8 March
%I Association for Computational Linguistics
%C Rabat, Morocco
%@ 979-8-89176-386-9
%F halder-etal-2026-riddlebench
%X While Large Language Models (LLMs) show remarkable capabilities, their complex reasoning skills require deeper investigation. We introduce **RiddleBench**, a new benchmark of 1,737 challenging puzzles designed to test reasoning beyond simple pattern matching. Our evaluation of state-of-the-art models reveals significant limitations, including hallucination cascades (uncritically accepting flawed peer reasoning) and poor self-correction due to strong self-confirmation bias. We also find that model performance is fragile, degrading when faced with reordered constraints or irrelevant information. RiddleBench serves as a resource for diagnosing these issues and guiding the development of more robust LLMs.
%U https://aclanthology.org/2026.findings-eacl.228/
%P 4363-4372

Download as File

Markdown (Informal)

[RiddleBench: A New Generative Reasoning Benchmark for LLMs](https://aclanthology.org/2026.findings-eacl.228/) (Halder et al., Findings 2026)

RiddleBench: A New Generative Reasoning Benchmark for LLMs (Halder et al., Findings 2026)

ACL

Deepon Halder, Alan Saji, Thanmay Jayakumar, Anoop Kunchukuttan, Ratish Puduppully, and Raj Dabre. 2026. RiddleBench: A New Generative Reasoning Benchmark for LLMs. In Findings of the Association for Computational Linguistics: EACL 2026, pages 4363–4372, Rabat, Morocco. Association for Computational Linguistics.