The exponential growth of scientific publications poses a formidable challenge for researchers seeking to validate emerging hypotheses or synthesize existing evidence. In this paper, we introduce Valsci, an open-source, self-hostable utility that automates large-batch scientific claim verification using any OpenAI-compatible large language model. Valsci unites retrieval-augmented generation with structured bibliometric scoring and chain-of-thought prompting, enabling users to efficiently search, evaluate, and summarize evidence from the Semantic Scholar database and other academic sources. Unlike conventional standalone LLMs, which often suffer from hallucinations and unreliable citations, Valsci grounds its analyses in verifiable published findings. A guided prompt-flow approach is employed to generate query expansions, retrieve relevant excerpts, and synthesize coherent, evidence-based reports. Preliminary evaluations across claims from the SciFact benchmark dataset reveal that Valsci significantly outperforms base GPT-4o outputs in citation hallucination rate while maintaining a low misclassification rate. The system is highly scalable, processing hundreds of claims per hour through asynchronous parallelization. By providing an open and transparent platform for large-batch literature verification, Valsci substantially lowers the barrier to comprehensive evidence-based reviews and fosters a more reproducible research ecosystem.
DeepSeek is a free and self-hostable large language model (LLM) that recently became the most downloaded app across 156 countries. As early academic literature on ChatGPT was predominantly critical of the model, this mini-review is interested in examining how DeepSeek is being evaluated across academic disciplines. The review analyzes available articles with DeepSeek in the title, abstract, or keywords, using the VADER sentiment analysis library. Due to limitations in comparing sentiment across languages, we excluded Chinese literature in our selection. We found that Computer Science, Engineering, and Medicine are the most prominent fields studying DeepSeek, showing an overall positive sentiment. Notably, Computer Science had the highest mean sentiment and the most positive articles. Other fields of interest included Mathematics, Business, and Environmental Science. While there is substantial academic interest in DeepSeek's practicality and performance, discussions on its political or ethical implications are limited in academic literature. In contrast to ChatGPT, where all early literature carried a negative sentiment, DeepSeek literature is mainly positive. This study enhances our understanding of DeepSeek's reception in the scientific community and suggests that further research could explore regional perspectives.
As the amount of genomic data continues to grow, there is an increasing need for systematic ways to organize, explore, compare, analyze and share this data. Despite this, there is a lack of suitable platforms to meet this need. OpenGenomeBrowser is a self-hostable, open-source platform to manage access to genomic data and drastically simplifying comparative genomics analyses. It enables users to interactively generate phylogenetic trees, compare gene loci, browse biochemical pathways, perform gene trait matching, create dot plots, execute BLAST searches, and access the data. It features a flexible user management system, and its modular folder structure enables the organization of genomic data and metadata, and to automate analyses. We tested OpenGenomeBrowser with bacterial, archaeal and yeast genomes. We provide a docker container to make installation and hosting simple. The source code, documentation, tutorials for OpenGenomeBrowser are available at opengenomebrowser.github.io and a demo server is freely accessible at opengenomebrowser.bioinformatics.unibe.ch . To our knowledge, OpenGenomeBrowser is the first self-hostable, database-independent comparative genome browser. It drastically simplifies commonly used bioinformatics workflows and enables convenient as well as fast data exploration.
We introduce CFAgentBench, a reproducible, self-hostable environment and benchmark for autonomous construction-finance agents: a CFO/controller-class agent operating across the real software stack a US construction finance team runs - ERP, project management, email, documents, pay applications, payroll, certified payroll, lien waivers, and bank/treasury portals. It contains 1,014 machine-gradeable task specifications across 8 domains and 77 families, every family grounded in a real source; a self-validated subset of 40 tasks (54 with a project-management extension) is compiled into oracle-validated executable evaluators, the runnable suite reported here. Following WebArena, the benchmark runs on an executable environment rather than static traces: 35 mock applications (31 reconciled to one company book, plus 4 PM platforms) over 9 archetypes, each implementing a uniform self-hostable app contract, so every task is graded by functional correctness - a state diff plus forbidden-side-effect checks plus required-output regexes - with an LLM judge used only for reply quality, never as reward. A distinguishing principle is a money-movement guard: 278 instances embed a payment, payroll, e
Environmental scientists spend disproportionate effort on data wrangling rather than analysis, and AI agents that automate geospatial workflows remain unvalidated: no benchmark evaluates agents operating through structured tool calling against real APIs. We introduce the GeoNatureAgent Benchmark, the first benchmark for environmental analysis agents that operate via structured tool calls to a production-style geospatial API. It comprises 93 tasks across 18 categories, covering municipality analysis, multi-turn conversation, spatial reasoning, cross-indicator synthesis, error handling and recovery, ranking, comparison, multilingual understanding, habitat analysis, and task rejection. Tasks are evaluated against an open, self-hostable API serving three environmental indicators across Spain and Portugal via sixteen tools. We evaluate seven LLMs (Claude Sonnet 4, DeepSeek V3.2, GLM-5, Gemini 2.5 Pro, Qwen3-235B, GPT-OSS-120B, Llama 4 Scout) under three temperature-1.0 seeds, reporting capability and per-case cost as orthogonal axes. We find: (1) Claude Sonnet 4 leads at 60.8% +/- 0.8%, followed by DeepSeek V3.2 at 56.3% +/- 3.1%, with no other model above 51%; (2) the cost-accuracy Par
We present a technical tutorial for building enterprise-grade realtime voice agents from first principles. While end-to-end speech-to-speech models may ultimately provide the best latency for voice agents, fully self-hosted end-to-end solutions are not yet available. We evaluate the closest candidate, Qwen3-Omni, across three configurations: its cloud-only DashScope Realtime API achieves $\sim$702ms audio-to-audio latency with streaming, but is not self-hostable; its local vLLM deployment supports only the Thinker (text generation from audio, 516ms), not the Talker (audio synthesis); and its local Transformers deployment runs the full pipeline but at $\sim$146s -- far too slow for realtime. The cascaded streaming pipeline (STT $\rightarrow$ LLM $\rightarrow$ TTS) therefore remains the practical architecture for self-hosted realtime voice agents, and the focus of this tutorial. We build a complete voice agent using Deepgram (streaming STT), vLLM-served LLMs with function calling (streaming text generation), and ElevenLabs (streaming TTS), achieving a measured time-to-first-audio of 755ms (best case 729ms) with full function calling support. We release the full codebase as a 9-chapte
Text embeddings for Portuguese have no dedicated benchmark: evaluation rests on translated corpora such as English MS MARCO or on thin multilingual coverage, with native tasks scattered and unconsolidated. We introduce MTEB-BR, a benchmark of 22 native Brazilian-Portuguese tasks across seven categories (classification, multilabel classification, pair classification, semantic textual similarity, clustering, retrieval, and reranking), admitting only data created or found in Portuguese and excluding translations by construction. We evaluate 93 models spanning 23M to 27B parameters: 73 open-weight and 20 closed commercial APIs. Alongside the leaderboard we report a statistical layer for every headline comparison: per-task bootstrap confidence intervals, paired-bootstrap significance, a task- and instance-level discrimination analysis (how sharply each task separates models) adapted from Item Response Theory, and a cross-leaderboard correlation. Three findings stand out. The benchmark cleanly separates about a dozen tiers of models, though the top six are statistically too close to order. An openly licensed, self-hostable model reaches that leading tier, so strong Portuguese embedding q
ResearchPilot is an open-source, self-hostable multi-agent system for literature-review assistance. Given a natural-language research question, it retrieves papers from Semantic Scholar and arXiv, extracts structured findings from paper abstracts, synthesizes cross-paper patterns, and drafts a citation-aware related-work section. The system combines FastAPI, Next.js, DSPy, SQLite, and Qdrant in a local-first architecture that supports bring-your-own-key model access and remote-or-local embeddings. This paper describes the system design, typed agent interfaces, persistence and history-search mechanisms, and the engineering tradeoffs involved in building a transparent research assistant. Rather than claiming algorithmic novelty, we present ResearchPilot as a systems contribution and evaluate it through automated tests and end-to-end local runs. We discuss limitations including external API rate limits, abstract-only extraction, incomplete corpus coverage, and the lack of citation verification.
Formal languages are an integral part of modeling and simulation. They allow the distillation of knowledge into concise simulation models amenable to automatic execution, interpretation, and analysis. However, the arguably most humanly accessible means of expressing models is through natural language, which is not easily interpretable by computers. Here, we evaluate how a Large Language Model (LLM) might be used for formalizing natural language into simulation models. Existing studies only explored using very large LLMs, like the commercial GPT models, without fine-tuning model weights. To close this gap, we show how an open-weights, 7B-parameter Mistral model can be fine-tuned to translate natural language descriptions to reaction network models in a domain-specific language, offering a self-hostable, compute-efficient, and memory efficient alternative. To this end, we develop a synthetic data generator to serve as the basis for fine-tuning and evaluation. Our quantitative evaluation shows that our fine-tuned Mistral model can recover the ground truth simulation model in up to 84.5% of cases. In addition, our small-scale user study demonstrates the model's practical potential for
Ronald Fischer, aka Richard Graydon, was arrested in New York last week
Anthropic blocks authors from opting out of $1。5B settlement at last minute
NatGeo's Pompeii: Out of Time fuses historical fact and imagination to bring city's last 24 hours to life
"Current economic conditions" have shifted TreeSize's business model
It turns out some of Benjamin Franklin's opinions from 240 years ago are still relevant today
Researchers have achieved a major milestone by creating a long-sought two-dimensional quantum material and confirming its unusual conducting edge states。 The ability to control these states through strain could make the material a promising platform for future room-temperature quantum electronics
Researchers have created cosmic dust from scratch by recreating space-like conditions inside glass tubes。 The dust contains complex carbon-rich molecules built from elements essential to life and produces infrared signals similar to real material found in space。 By studying these laboratory samples, scientists can explore how organic chemistry unfo
Physicists from Heinrich Heine University Düsseldorf (HHU) have examined a fundamental property of quantum mechanics in collaboration with the German Aerospace Center (DLR)。 In the scientific journal Physical Review Letters, they show that this theory does not necessarily need to be formulated with imaginary numbers – real numbers can in fact also
Scientists say new technologies have reopened the debate over whether Mars could someday be terraformed, turning a once impossible idea into a serious research topic。 Before anyone tries to reshape the Red Planet, though, researchers say we must understand the risks, including what might be lost if Mars already harbors its own forms of life
A catastrophic asteroid breakup may have triggered a huge wave of impacts across the inner solar system about 800 million years ago。 The debris was launched from near a gravitational gateway controlled by Jupiter, sending fragments toward Earth, the Moon, and Mars。 The bombardment may explain ancient lunar craters and could have contributed to majo
A mirror that moves mid-photon could release a shower of new photons