i love that the team does stuff like this
i love that the team does stuff like this Codex is not like claude code. if you know the limit is going to end, like last 10 to 8%, give an very long run task, and even after the limit got ever, it wi
Knowledge catalogue
i love that the team does stuff like this Codex is not like claude code. if you know the limit is going to end, like last 10 to 8%, give an very long run task, and even after the limit got ever, it wi
I released LLM 0.32a0 this morning, a major backwards-compatible refactor of my LLM Python library and CLI tool for working with language models - the new changes should help LLM work better with reas
IBM Granite just released two multilingual embedding models with 97M and 311M parameters 🤏🏻 ModernBERT-based, 200+ languages, 32K context, and built for retrieval, search, similarity, and code. And...
arXiv:2604.25188v1 Announce Type: new Abstract: Image classification remains a fundamental yet challenging task in computer vision, particularly when fine-grained feature extraction and background noi
arXiv:2505.13302v2 Announce Type: replace Abstract: As language and vision-language models (VLMs) become central to information access and online interaction, concerns grow about their potential to am
IMO DeepSeek v4 demonstrated utter confidence and competence by not benchmaxxing, not focusing on some BS final run cost, not even spending inference-optimal compute. just showed up, demonstrated SOTA
arXiv:2603.22665v2 Announce Type: replace Abstract: The standard practice in Large Language Models (LLMs) is to base predictions on final-layer representations. However, intermediate layers encode com
arXiv:2604.24827v1 Announce Type: new Abstract: Closed-source frontier labs do not disclose parameter counts, and the standard alternative -- inference economics -- carries 2imes+ uncertainty from har
arXiv:2508.02964v3 Announce Type: replace Abstract: Diffusion models have been firmly established as principled zero-shot solvers for linear and nonlinear inverse problems, owing to their powerful ima
arXiv:2604.24765v1 Announce Type: cross Abstract: P300-based brain-computer interfaces (BCIs) are widely used for communication, but population heterogeneity may alter the neural patterns available fo
🚀 Introducing FlashQLA: high-performance linear attention kernels built on TileLang. ⚡ 2–3× forward speedup. 2× backward speedup. 💻 Purpose-built for agentic AI on your personal devices. 💡Key insights
Introducing remote agents in Vibe and Mistral Medium 3.5. You can now launch remote agents in the cloud, including from the CLI or Le Chat. Plus, new Work mode in Le Chat for complex, multi-step tasks
arXiv:2604.25858v1 Announce Type: new Abstract: Transformers have demonstrated a strong ability for in-context learning (ICL), enabling models to solve previously unseen tasks using only example input
arXiv:2507.12553v3 Announce Type: replace Abstract: Language models (LMs) are used for a diverse range of tasks, from question answering to writing fantastical stories. In order to reliably accomplish
Its a bit frustrating, because Gemini 3.1 Pro is an excellent model and can deliver really good results. But here is GPT-5.5 Pro for comparison. Sadly, it took this assignment very seriously and ethic
arXiv:2602.15547v2 Announce Type: replace Abstract: Text embedding models are widely used for semantic similarity tasks, including information retrieval, clustering, and classification. General-purpos
arXiv:2604.25788v1 Announce Type: new Abstract: Robotic systems that interact with the physical world must reason about kinematic and dynamic constraints imposed by their own embodiment, their environ
arXiv:2604.24927v1 Announce Type: new Abstract: Generating diverse responses is crucial for test-time scaling of large language models (LLMs), yet standard stochastic sampling mostly yields surface-le
Last week we announced DeepSeek-V4. Today we’re sharing a closer look at DeepSeek-V4 Pro on Together AI: 512K context, controllable reasoning modes, and cached-input pricing for long-context workloads
arXiv:2604.25178v1 Announce Type: new Abstract: Achieving a desirable balance between rendering quality and real-time performance is a long-standing challenge in modern game and rendering engines, par
arXiv:2604.24788v1 Announce Type: new Abstract: Natural gas is undoubtedly an essential component of the global energy system. Accurate short-term forecasting of natural gas price is challenging due t
AI21 Labs announced that their Chief Product Officer Or Dagan presented at the DeepLearning.AI conference, discussing how the company achieved state-of-the-art results on the Browsecomp-Plus benchmark
I just released LLM 0.32a0, an alpha release of my LLM Python library and CLI tool for accessing LLMs, with some consequential changes that I've been working towards for quite a while. Previous versio
arXiv:2604.25665v1 Announce Type: new Abstract: Reliable evaluation of large language model (LLM)-generated summaries remains an open challenge, particularly across heterogeneous domains and document
arXiv:2604.25130v1 Announce Type: new Abstract: Evaluating long document summaries remains the primary bottleneck in summarization research. Existing metrics correlate weakly with human judgments and
Cybersecurity company Lookout Inc. today announced the launch of Lookout AI Visibility & Governance, a new mobile-native solution designed to provide organizations with the visibility needed to discov
arXiv:2604.25122v1 Announce Type: new Abstract: We present M^3-VQA, a novel knowledge-based Visual Question Answering (VQA) benchmark, to enhance the evaluation of multimodal large language models (ML
arXiv:2601.21225v2 Announce Type: replace Abstract: Large language models have made substantial progress in mathematical reasoning. However, benchmark development for multilingual evaluation has lagge
arXiv:2512.07348v2 Announce Type: replace Abstract: In controllable image generation, synthesizing coherent and consistent images from multiple reference inputs, i.e., Multi-Image Composition (MICo),
arXiv:2604.24805v1 Announce Type: new Abstract: Modern machine learning optimizes for accuracy without explicitly accounting for internal computational cost, even though physical and biological system
arXiv:2603.19874v3 Announce Type: replace-cross Abstract: Loss functions play a central role in supervised classification. Cross-entropy (CE) is widely used, whereas the mean absolute error (MAE) loss
arXiv:2510.22102v2 Announce Type: replace-cross Abstract: While Multimodal Large Language Models (MLLMs) excel at general vision-language tasks, precise coordinate prediction remains a significant cha
arXiv:2512.17492v2 Announce Type: replace Abstract: Geo-spatial analysis of our world benefits from a multimodal approach, as every single geographic location can be described in numerous ways (images
Model-Harness-Task fit is very real. The world's best agents carefully tailor the harness around the model to take advantage of each model's unique intelligence & capabilities 🧠 Today we released Harn
arXiv:2604.24809v1 Announce Type: new Abstract: We present Nautile-370M, a 371-million-parameter small language model designed for efficient reasoning under strict parameter and inference budgets. Nau
arXiv:2604.24954v1 Announce Type: cross Abstract: We introduce Nemotron 3 Nano Omni, the latest model in the Nemotron multimodal series and the first to natively support audio inputs alongside text, i
arXiv:2305.06709v4 Announce Type: replace Abstract: NUBO, short for Newcastle University Bayesian Optimisation, is a Bayesian optimization framework for the optimization of expensive-to-evaluate black
arXiv:2604.24964v1 Announce Type: cross Abstract: Existing web agent benchmarks have largely converged on short, single-site tasks that frontier models are approaching saturation on. However, real wor
arXiv:2508.16198v3 Announce Type: replace Abstract: Multimodal Large Language Models (MLLMs) have increasingly supported omni-modal processing across text, vision, and speech. However, existing evalua
arXiv:2604.25102v1 Announce Type: new Abstract: Typographic prompt injection exploits vision language models' (VLMs) ability to read text rendered in images, posing a growing threat as VLMs power auto
arXiv:2512.03043v3 Announce Type: replace Abstract: Reinforcement learning (RL) has recently achieved remarkable success in eliciting visual reasoning within Multimodal Large Language Models (MLLMs).
OpenAI announced the return of DevDay, their developer conference, scheduled for September 29 in San Francisco. The event typically features product announcements, demonstrations of new capabilities,
arXiv:2412.00167v2 Announce Type: replace Abstract: In recent years, origin-destination (OD) demand prediction has gained significant attention for its profound implications in urban development. Exis
Seema Mody / CNBC: Pentagon's Digital and AI Chief Cameron Stanley confirms the DoD is expanding its Google Gemini use, saying “overreliance on one vendor is never a good thing” — Pentagon AI chief Ca
arXiv:2502.02452v4 Announce Type: replace Abstract: Personalization of Large Vision-Language Models (LVLMs) involves customizing models to recognize specific users or object instances and to generate
arXiv:2604.05030v2 Announce Type: replace Abstract: Experiments probing natural language processing by both humans and LLMs suggest that the meaning of a semantic expression is indeterminate prior to
arXiv:2604.24971v1 Announce Type: cross Abstract: We present PolyKV, a system in which multiple concurrent inference agents share a single, asymmetrically compressed KV cache pool. Rather than allocat
arXiv:2410.04182v2 Announce Type: replace Abstract: While portrait sketch generation is a special task in sketch synthesis, most existing methods are pixel-based, limiting their interpretability and e
arXiv:2604.25441v1 Announce Type: cross Abstract: Commercial TTS systems produce near-native Indic audio, but the best open-source bases (Chatterbox, Indic Parler-TTS, IndicF5) trail them on measured
arXiv:2604.25154v1 Announce Type: new Abstract: Tabular Foundation Models (TFMs) achieve state-of-the-art zero-shot accuracy on small tabular datasets by meta-learning over synthetic data-generating p
Proactive agent that thinks and acts like you. Multiplayer AI Brain for teams. Proper GUI for commanding 50 agents. http://Sauna.ai is all three. Sauna goes live today. First 2000 people, use access c
arXiv:2604.25840v1 Announce Type: new Abstract: Patient simulators are gaining traction in mental health training by providing scalable exposure to complex and sensitive patient interactions. Simulati
arXiv:2604.25476v1 Announce Type: cross Abstract: Standard text-to-speech (TTS) evaluation measures intelligibility (WER, CER) and overall naturalness (MOS, UTMOS) but does not quantify accent. A synt
arXiv:2604.25688v1 Announce Type: new Abstract: Binary spike coding enables sparse and event-driven computation in spiking neural networks (SNNs), yet its 1-bit-per-timestep representation fundamental
arXiv:2604.25884v1 Announce Type: cross Abstract: Quantum computing calibration depends on interpreting experimental data, and calibration plots provide the most universal human-readable representatio
arXiv:2602.17262v2 Announce Type: replace Abstract: Human self-report questionnaires are increasingly used in NLP to benchmark and audit large language models (LLMs), from persona consistency to safet
arXiv:2604.24803v1 Announce Type: new Abstract: In low-depth implementations of the Quantum Approximate Optimization Algorithm (QAOA), the dominant cost is often the number of objective evaluations ra
Qwen 3.5 from @Alibaba_Qwen is now available on @FireworksAI_HQ Training Platform across the Managed and Training API workflows. Try SFT, DPO, RL with smart defaults or your own custom loss function w
FlashQLA is a high-performance linear attention kernel library built on TileLang developed by Alibaba's Qwen team. The introduction of FlashQLA represents an optimization technology designed to improv
arXiv:2603.09723v2 Announce Type: replace Abstract: Large language models (LLMs) are increasingly used across the scientific workflow, including to draft peer-review reports. However, many AI-generate