Papersby GlyphStrategic Navigation or Stochastic Search? How Agents and Humans Reason Over Document Collections research paper by Snowflake, 2026
Snowflake · Agents and evaluation · Mar 12, 2026 · 3 citations · 66 upvotes
What it shows
On 2,250 questions over 800 PDFs, the best agents match human accuracy but lean on brute-force search rather than planning.
By Łukasz Borchmann, Jordy Van Landeghem, Michał Turski and 12 more · arXiv 2603.12180 · PDF · Code
Summarised by hand from the abstract.
More from Snowflake
All 5Agent World Model: Infinity Synthetic Environments for Agentic Reinforcement LearningLarge language model agents trained in synthetic environments with code-driven simulations and database-backed state transitions demonstrate superior out-of-distribution generalization compared to traditional benchmark-specific approaches.
Snowflake81 citations53 upvotesRead →Language Models Model LanguageThe paper advocates for an empiricist approach to evaluating language models, emphasizing frequency of use over traditional theoretical frameworks.Snowflake26 upvotesRead →SwiftKV: Fast Prefill-Optimized Inference with Knowledge-Preserving Model TransformationSwiftKV lets prompt tokens skip later layers and merges KV caches, cutting prefill cost for long-prompt workloads.Snowflake21 citations2 upvotesRead →Arctic-Embed: Scalable, Efficient, and Accurate Text Embedding ModelsOpen text embedding models from 22M to 334M parameters that led MTEB retrieval for their size at release.Snowflake84 citations2 upvotesRead →
Other agents and evaluation papers
TopicGemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic CapabilitiesGemini 2.X model family, including Gemini 2.5 Pro and Flash, offers superior coding, reasoning, and multimodal understanding capabilities across a range of computational efficiencies.
Google4,594 citations67 upvotesRead →DeepSeek-V3.2: Pushing the Frontier of Open Large Language ModelsDeepSeek-V3.2 introduces DeepSeek Sparse Attention and a scalable reinforcement learning framework, achieving superior reasoning and performance compared to GPT-5 and Gemini-3.0-Pro in complex reasoning tasks.
DeepSeek887 citations275 upvotesRead →WorkArena: How Capable Are Web Agents at Solving Common Knowledge Work Tasks?WorkArena and BrowserGym evaluate large language model-based agents' ability to perform enterprise software tasks, revealing gaps in current agent capabilities and differences between open and closed-source LLMs.
ServiceNow443 citations2 upvotesRead →Aya Model: An Instruction Finetuned Open-Access Multilingual Language ModelAya, a multilingual generative language model supporting over 50% lower-resourced languages, excels in both generative and discriminative tasks across 99 languages, and provides extensive evaluations and open-source resources.
Cohere438 citations48 upvotesRead →Rubrics as Rewards: Reinforcement Learning Beyond Verifiable DomainsRubrics as Rewards (RaR) framework uses structured rubrics as interpretable reward signals for on-policy training, improving performance in real-world reinforcement learning tasks with subjective criteria.
Scale AI431 citations6 upvotesRead →SWE-Bench Pro: Can AI Agents Solve Long-Horizon Software Engineering Tasks?SWE-Bench Pro is a challenging benchmark for coding models, featuring complex, enterprise-level problems that require substantial code modifications, with performance evaluations showing significant limitations in current models.Scale AI367 citations21 upvotesRead →
About this paper
- Venue
- arXiv.org
- Citations
- 3, 2 influential · Semantic Scholar
- Upvotes
- 66 · Hugging Face
- Lab
- Snowflake · on Companies · on Acquisitions · on Quarterly · on Paydays · on Releases · on TechConf
Changes
Influential citations rose from 1 to 2.
Citations rose from 2 to 3.
Influential citations rose from 0 to 1.
Citations rose from 1 to 2.
Added to the list.