Knowledge catalogue
Today we open sourced many of OpenAI's monitorability evaluations. We hope that the research community and other model developers can build upon them and use them to evaluate the monitorability of the
arXiv:2604.21534v1 Announce Type: new Abstract: This paper presents our system developed for SemEval-2026 Task 2. The task requires modeling both current affect and short-term affective change in chro
arXiv:2504.02181v2 Announce Type: replace Abstract: The rapid advancements in large Language models (LLMs) have significantly enhanced their reasoning capabilities, driven by various strategies such a
A developer describes building a persistent-state AI system using Ollama with three models (a 4B router model, a 9B primary model, and a 9B 'subconscious' model) running on dual RTX 3060 GPUs, demonst
arXiv:2604.20666v1 Announce Type: cross Abstract: Effective retrieval-augmented generation across bilingual Greek--English applications requires embedding models capable of capturing both domain-speci
arXiv:2604.20000v1 Announce Type: new Abstract: Automated wildlife monitoring from aerial imagery is vital for conservation but remains limited by two persistent challenges: the difficulty of detectin
arXiv:2604.19838v1 Announce Type: new Abstract: Understanding how road users resolve space-sharing conflicts is important both for traffic safety and the safe deployment of autonomous vehicles. While
arXiv:2604.01577v2 Announce Type: replace-cross Abstract: We extend the recent latent recurrent modeling to sequential input streams. By interleaving fast, recurrent latent updates with self-organizat
arXiv:2604.19850v1 Announce Type: cross Abstract: Biological systems are promising substrates for computation because they naturally process environmental information through complex internal dynamics
zero shot Kimi K2.6, go try it out its a good model sir! this is @Kimi_Moonshot running on @togethercompute, @opencode harness prompt below👇 Media Introducing Kimi K2.6 from @Kimi_Moonshot, a multimod
GPT Image 2.0 just dropped in ComfyUI via Partner Nodes. This isn't another image model. It *reasons* before it generates. → Plans the composition → Checks its own work → Iterates instead of one-shott
Imagine every pixel on your screen, streamed live directly from a model. No HTML, no layout engine, no code. Just exactly what you want to see. @eddiejiao_obj, @drewocarr and I built a prototype to se
Kimi K2.6, a model developed by Moonshot AI, has achieved the top ranking on OpenRouter's programming leaderboard, indicating superior performance in code generation and programming-related tasks comp
Perplexity's reward design system prioritizes correctness as a foundational requirement, then evaluates user preference and efficiency only when answers meet correctness standards. This hierarchical a
At Google Cloud Next, we are announcing innovations across every layer of our storage stacks — performance, intelligence, and management — to ensure your data is as fast and as useful as the AI models
This post highlights the Qwen 3.6-27B language model's image generation capabilities, noting that despite its relatively compact 16.8GB file size, it produces high-quality creative outputs like the ex
arXiv:2506.07969v2 Announce Type: replace Abstract: We consider the problem of modeling high-speed flows using machine learning methods. While most prior studies focus on low-speed fluid flows in whic
arXiv:2604.16733v1 Announce Type: new Abstract: Physical awareness, especially in a large and dynamic environment, is shaped by sensing decisions that determine observability across space, time, and s
arXiv:2604.17085v1 Announce Type: new Abstract: The interpretation of implicit meanings is an integral aspect of human communication. However, this framework may not transfer to interactions with Larg
arXiv:2604.17074v1 Announce Type: new Abstract: The rapid advancement of generative models has led to a growing volume of AI-generated videos, making the automatic quality assessment of such videos in
arXiv:2604.17833v1 Announce Type: new Abstract: What appears effortless to a human waiter remains a major challenge for robots. Manipulating objects nonprehensilely on a tray is inherently difficult,
arXiv:2604.16484v1 Announce Type: new Abstract: Deploying generative World-Action Models for manipulation is severely bottlenecked by redundant pixel-level reconstruction, O(T) memory scaling, and seq
arXiv:2604.18563v1 Announce Type: new Abstract: A recent study (Kuribayashi et al., 2025) has shown that human sentence processing behavior, typically measured on syntactically unchallenging construct
arXiv:2604.18106v1 Announce Type: new Abstract: Adapting large language models (LLMs) to low-resource languages (LRLs) is constrained by the scarcity of task data and computational resources. Although
Kimi 2.6 Thinking seems very good for an open weights model, but many rough edges compared to closed SoTA. The Lem Test resulted in a 74 page thinking trace... and an okay-ish answer. It did an okay T
Kimi K2.6, a language model developed by Moonshot AI, has achieved the top ranking on the open-weight category of the Vals Index benchmark, while placing 7th overall across all model categories. The V
arXiv:2509.11983v2 Announce Type: replace Abstract: Neural network (NN) training is inherently a large-scale matrix optimization problem, yet the matrix structure of NN parameters has long been overlo
This DigitalOcean guide covers strategies and best practices for deploying and optimizing very large language models (600 billion+ parameters) on cloud infrastructure, focusing on inference performanc
arXiv:2604.18069v1 Announce Type: new Abstract: Humans often hold different perspectives on the same issues. In many NLP tasks, annotation disagreement can reflect valid subjective perspectives. Model
arXiv:2604.17972v1 Announce Type: new Abstract: Emotional Support Conversation (ESC) aims to assist individuals experiencing distress by generating empathetic and supportive dialogue. While prior work
arXiv:2604.16967v1 Announce Type: new Abstract: Path planning is usually solved by addressing either the (high-level) route planning problem (waypoint sequencing to achieve the final goal) or the (low
arXiv:2604.18002v1 Announce Type: new Abstract: Chain-of-thought reasoning has driven striking advances in language model capability, yet every reasoning step grows the KV cache, creating a bottleneck
arXiv:2604.17338v1 Announce Type: cross Abstract: Unlike code completion, debugging requires localizing faults and applying targeted edits. We observe that frontier LLMs often regenerate correct but o
OpenAI has been using multiple informal names internally and externally for its new image generation model, including 'ChatGPT Images 2.0,' 'Image gen 2,' and 'gpt-image-2.' The post highlights incons
arXiv:2604.18484v1 Announce Type: new Abstract: Vision-Language-Action (VLA) models drive next-generation autonomous systems, but training them requires scalable, high-quality annotations from complex
Anthropic announced its Mythos AI model on April 7 and refused to release it publicly, citing its unprecedented ability to find and exploit software vulnerabilities . The company is instead limiting a
arXiv:2602.01766v2 Announce Type: replace-cross Abstract: The quadratic complexity and indefinitely growing key-value (KV) cache of standard Transformers pose a major barrier to long-context processin
arXiv:2604.15866v1 Announce Type: cross Abstract: Large language models (LLMs) have advanced information extraction (IE) by enabling zero-shot and few-shot named entity recognition (NER), yet their ge
arXiv:2509.14977v2 Announce Type: replace Abstract: Ultrasound imaging has become the preferred imaging modality for early cancer screening due to its advantages of non-ionizing radiation, low cost, a
Kimi has been the most popular model on Fireworks, both out of the box and as a fine-tuning base (including Composer 2) Now Kimi K2.6 is live with huge jumps (10+%) in coding, long-running agents and
arXiv:2604.15760v1 Announce Type: new Abstract: We introduce the first version of KWBench (Knowledge Work Bench), a benchmark for unprompted problem recognition in large language models: can an LLM id
literally my basic model since 1998. crazy that some people still haven’t figured this out. My basic model of capabilities: LLMs are good at problems similar to those that appear in their training dat
arXiv:2604.15314v1 Announce Type: cross Abstract: This research aimed to develop an intelligent system to evaluate performance and extract behavioral models for children with ASD and neurotypical (TD)
Nice paper combining the strength of Skills and RAG. Most RAG systems retrieve on every query, whether the model needs help or not. This is wasteful when the model already knows the answer, and often
arXiv:2604.03444v3 Announce Type: replace-cross Abstract: Recent work has demonstrated the potential of non-transformer language models, especially linear recurrent neural networks (RNNs) and hybrid m
opus 4.7 dropped last week with two new features: xhigh effort: more compute when you need it task budgets: the model knows its limit and wraps up gracefully, this is a great example of giving the mod
Our researchers are heading to ICLR with new work: model efficiency, long-context reasoning, next-gen attention and decoding, and more. Check out what we've been building 👇🏼 #TogetherResearch #AINativ
arXiv:2512.06987v2 Announce Type: replace Abstract: Accurately predicting experimentally realizable 3D molecular crystal structures from their 2D chemical graphs is a long-standing open challenge in c
arXiv:2604.15483v1 Announce Type: new Abstract: We present a new robotic foundation model, called {pi}_{0.7}, that can enable strong out-of-the-box performance in a wide range of scenarios. {pi}_{0.7}
arXiv:2604.15670v1 Announce Type: new Abstract: Reasoning segmentation has recently expanded from ground-level scenes to remote-sensing imagery, yet UAV data poses distinct challenges, including obliq
arXiv:2604.15780v1 Announce Type: cross Abstract: Machine learning models are increasingly deployed in real-world applications, but even aligned models such as Mistral and LLaVA still exhibit unsafe b
arXiv:2602.05638v3 Announce Type: replace Abstract: While foundation models have advanced surgical video analysis, current approaches rely predominantly on pixel-level reconstruction objectives that w
arXiv:2604.13812v1 Announce Type: new Abstract: Quantum circuit optimization is a central task in Quantum Computing, as current Noisy Intermediate Scale Quantum devices suffer from error propagation t
arXiv:2510.24538v2 Announce Type: replace Abstract: Textual humor is enormously diverse and computational studies need to account for this range, including intentionally bad humor. In this paper, we c
arXiv:2604.13151v1 Announce Type: new Abstract: Language Model (LM) agents are increasingly used in complex open-ended decision-making tasks, from AI coding to physical AI. A core requirement in these
arXiv:2604.15109v1 Announce Type: new Abstract: Despite the rapid advancement of Large Language Models (LLMs), uncertainty quantification in LLM generation is a persistent challenge. Although recent a
arXiv:2512.05024v3 Announce Type: replace-cross Abstract: As generative AI models are increasingly used to simulate real-world systems, quantifying the ``sim-to-real'' gap is critical. For each input
arXiv:2604.14808v1 Announce Type: new Abstract: Machine unlearning for large language models (LLMs) aims to remove targeted knowledge while preserving general capability. In this paper, we recast LLM
arXiv:2604.14197v1 Announce Type: new Abstract: Large language model (LLM) performance depends heavily on prompt design, yet prompt construction is often described and applied inconsistently. Our purp
Luz Ding / Bloomberg: Alibaba's new Token Hub unit releases Happy Oyster, a new AI world model that can create 3D environments, interactive videos, films, video content, and games — Alibaba Group Hold