2026-09-02 04:00 UTCOriginal source3 min readUpdated: 2026-09-02 09:07 UTC

Long-Horizon State Tracking in LLMs: Executing MD5 through a Deep Sequence of Dependent Tool Calls

arXiv:2609.00012v1 Announce Type: new Abstract: Long-horizon tasks remain uncommon in large language model (LLM) evaluation, and for a reason: when each step depends on the last, per-step accuracy that looks excellent in isolation decays catastrophically, as errors cascade and the end-to-end failure probability grows sharply with length. Existing agentic benchmarks report end-to-end success but confound this state-tracking difficulty with instruction interpretation, give no control group that isolates it, and are vulnerable to shortcuts such as a hallucinated final answer, so they cannot say why a long run fails. Whether an LLM can carry exact intermediate state across many tool calls at all is itself not well established. We test this cleanly by having the model compute a cryptographic hash, MD5, step by step: a sequence of $196$ dependent tool calls over $64$ rounds while it carries four $32$-bit words $(a,b,c,d)$ in its own context from one call to the next. Interpretation is trivial and, because we implement MD5 from scratch (RFC~1321), we align every call to the ground-truth trace and check the digest to the bit, so any failure is pure bookkeeping. gpt-oss-120b, a mixture-of-experts model with only $\sim$5.5B active parameters per token, at temperature $0$ with a short fixed prompt, carries the full state across all $196$ calls and returns the correct digest on a majority of completed runs. In the strongest setting we replace every primitive tool with a second LLM, so a driver and a worker compute the whole hash from scratch with no exact-arithmetic oracle in the loop. Two ingredients decide success and neither changes the weights: keeping the model's own reasoning in its context each turn, and voting over a thinking-enabled worker to remove its modular-arithmetic slips. We localize the residual failures by origin, separating state-carrying from arithmetic and from serving.

SourcearXiv AIAuthor: Dheeraj Mohandas Pai, Lu Xian

-->

[Submitted on 2 Aug 2026]

Title:Long-Horizon State Tracking in LLMs: Executing MD5 through a Deep Sequence of Dependent Tool Calls

View a PDF of the paper titled Long-Horizon State Tracking in LLMs: Executing MD5 through a Deep Sequence of Dependent Tool Calls, by Dheeraj Mohandas Pai and 1 other authors

View PDF HTML (experimental)

Abstract:Long-horizon tasks remain uncommon in large language model (LLM) evaluation, and for a reason: when each step depends on the last, per-step accuracy that looks excellent in isolation decays catastrophically, as errors cascade and the end-to-end failure probability grows sharply with length. Existing agentic benchmarks report end-to-end success but confound this state-tracking difficulty with instruction interpretation, give no control group that isolates it, and are vulnerable to shortcuts such as a hallucinated final answer, so they cannot say why a long run fails. Whether an LLM can carry exact intermediate state across many tool calls at all is itself not well established. We test this cleanly by having the model compute a cryptographic hash, MD5, step by step: a sequence of $196$ dependent tool calls over $64$ rounds while it carries four $32$-bit words $(a,b,c,d)$ in its own context from one call to the next. Interpretation is trivial and, because we implement MD5 from scratch (RFC~1321), we align every call to the ground-truth trace and check the digest to the bit, so any failure is pure bookkeeping. gpt-oss-120b, a mixture-of-experts model with only $\sim$5.5B active parameters per token, at temperature $0$ with a short fixed prompt, carries the full state across all $196$ calls and returns the correct digest on a majority of completed runs. In the strongest setting we replace every primitive tool with a second LLM, so a driver and a worker compute the whole hash from scratch with no exact-arithmetic oracle in the loop. Two ingredients decide success and neither changes the weights: keeping the model's own reasoning in its context each turn, and voting over a thinking-enabled worker to remove its modular-arithmetic slips. We localize the residual failures by origin, separating state-carrying from arithmetic and from serving.

Comments: 6 Pages, 2 figures

Subjects:

Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Multiagent Systems (cs.MA)

Cite as: arXiv:2609.00012 [cs.AI]

(or arXiv:2609.00012v1 [cs.AI] for this version)

https://doi.org/10.48550/arXiv.2609.00012

arXiv-issued DOI via DataCite

Submission history

From: Dheeraj Mohandas Pai [view email] [v1] Sun, 2 Aug 2026 09:38:02 UTC (532 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled Long-Horizon State Tracking in LLMs: Executing MD5 through a Deep Sequence of Dependent Tool Calls, by Dheeraj Mohandas Pai and 1 other authors

View PDF

HTML (experimental)

TeX Source

view license

Current browse context:

cs.AI

new | recent | 2026-09

Change to browse by:

cs cs.CR cs.MA

References & Citations

NASA ADS

Google Scholar

Semantic Scholar

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)