Bacteriophages, often referred to as the dark matter of the biosphere, play a critical role in regulating microbial ecosystems and in antibiotic alternatives. Thus, accurate interpretation of their genomes holds significant scientific and practical value. While general-purpose Large Language Models (LLMs) excel at understanding biological texts, their ability to directly interpret raw nucleotide sequences and perform biological reasoning remains underexplored. To address this, we introduce PhageBench, the first benchmark designed to evaluate phage genome understanding by mirroring the workflow of bioinformatics experts. The dataset contains 5,600 high-quality samples covering five core tasks across three stages: Screening, Quality Control, and Phenotype Annotation. Our evaluation of eight LLMs reveals that general-purpose reasoning models significantly outperform random baselines in phage contig identification and host prediction, demonstrating promising potential for genomic understanding. However, they exhibit significant limitations in complex reasoning tasks involving long-range dependencies and fine-grained functional localization. These findings highlight the necessity of developing next-generation models with enhanced reasoning capabilities for biological sequences.

Large language models (LLMs) have undergone rapid growth, forming a heterogeneous ecosystem whose model relationships and capability profiles lack comprehensive documentation. Organizing such an ecosystem requires reusable model representations that are comparable across architectures while still reflecting how each model processes shared inputs. We propose ABLE (Attribution-Based Large-model Embedding), a training-free framework that represents each model in an attribution space. Given a shared probing corpus, ABLE computes feature attributions, aligns token-level attribution scores into a tokenizer-agnostic word-level representation, and projects the resulting attribution patterns into compact model embeddings. These embeddings summarize model-level input-dependence patterns and can be reused across downstream tasks. We construct ABLE representations for 239 LLMs and evaluate them on model organization and relationship analysis, model routing, and benchmark score prediction. Across these tasks, ABLE captures meaningful model-level structure, achieves stronger relationship prediction than parameter- and output-based baselines, supports competitive model routing, and predicts benchmark scores from frozen representations, while theoretical analysis further supports the stability of the representation.

Large Language Models (LLMs) are increasingly adopted as conversational assistants in genomics, where they are mainly used to reason over biological knowledge, annotations, and analysis outputs through natural language interfaces. However, existing benchmarks either focus on specialized DNA models trained for sequence prediction or evaluate biological knowledge using text-only questions, leaving the behavior of general-purpose LLMs when directly exposed to raw genome sequences underexplored. We introduce GenomeQA, a benchmark designed to provide a controlled evaluation setting for general-purpose LLMs on sequence-based genome inference tasks. GenomeQA comprises 5,200 samples drawn from multiple biological databases, with sequence lengths ranging from 6 to 1,000 base pairs (bp), spanning six task families: Enhancer and Promoter Identification, Splice Site Identification, Taxonomic Classification, Histone Mark Prediction, Transcription Factor Binding Site Prediction, and TF Motif Prediction. Across six frontier LLMs, we find that models often outperform random baselines, particularly on tasks driven by local sequence cues such as GC content and short motifs, while performance degrades on tasks that require more indirect or multi-step inference over sequence patterns. GenomeQA establishes a diagnostic benchmark for studying and improving the use of general-purpose LLMs on raw genomic sequences.