2026-08-20

Anyone tried them yet? Disclaimer: Not affiliated with ornith. I just surf huggingface for new models every 30m or so. I'm addicted.

Aloha! 🌺Introducing Ornith-1.5, a family of open-source LLMs spanning 9B Dense, 35B MoE, and 397B MoE, trained with self-improving strategies. It achieves state-of-the-art performance among open-source models of comparable size and delivers performance comparable to Claude Opus 4.8 across reasoning, agentic, and coding tasks: ✅Terminal-Bench 2.1 (86.1) ✅SWE-Bench (86 on verified, 65.1 on pro, 79.6 on Multilingual) ✅DeepSWE (56) ✅HLE (44.6) ✅ClawEval (81.4) ✅Tool Decathlon (71.2)

llama.cpp pr #27342 adds dflash2, so i rented an rtx 6000 and ran the same four prompts through four decoding setups on qwen3.8 27B median results over the four tasks: baseline 47.4 tok/s mtp 114.7 tok/s dflash 99.3 tok/s dflash2 140.6. tok/s so on average 3x for dflash2 though i have to point out that it's far from a 3x gain some of the time, on one of the test it struggled to achieve a 1.5x gain, it really just depends on the task you give to the model the races are sped up in some places, so

Memory is becoming core infrastructure for long-horizon LLM agents, yet existing evaluations offer limited guidance on which memory substrate, namely the underlying medium in which memory is represented and stored, should be used under different operating regimes. We present a controlled harness evaluation of memory substrates for memory-augmented agents, covering dense and sparse indices, text records, structural stores, hierarchical stores, refinement-based memories, parametric updates, and ac

The Git forge built for the age of coding agents Discussion | Link

A competition is developing between OpenAI and Anthropic over who can provide the best privacy protections for enterprise customer data.

This goes way, way further than keeping an eye on license plates.

The U.S. phone provider escaped a large-scale breach of its network after identifying Chinese-backed hackers early on.

looks like GGUF files were just updated

the Memory crunch continues - Moore’s Law reversed to 2007 levels

arXiv:2608.16956v1 Announce Type: new Abstract: API buyers purchase a dated contract, not a model name alone: the contract includes the requested and served model, reasoning-effort term or its omission, output rail, service product, prompt, and price schedule. We study the reasoning-effort term through a registered paired contrast of Sonnet 5 with explicit high effort against the same model with effort omitted, using 30 AIME 2026 items and five calls per item. Every paid attempt was assigned one

arXiv:2608.16890v1 Announce Type: new Abstract: Clinical trial programming -- transforming study protocols into analysis-ready datasets under CDISC standards -- is a bottleneck in regulatory submissions, yet LLM-based code generation fails catastrophically on this task: across 11 single-shot attempts with five frontier models, none produces a valid subject-level analysis dataset. We introduce GxP-Agent, a multi-agent system that encodes regulatory process ordering as a directed acyclic graph (DA

arXiv:2608.17202v1 Announce Type: new Abstract: Safety alignment in open-weight language models is trivially removable: abliteration projects a refusal-mediating direction out of the weights in minutes, and no release-time defense we are aware of prevents it durably. What cannot be prevented can be deceived. Our defense, decoy hardening ("Fool's Gold"), concedes the refusal strip and poisons its payoff: once refusal is stripped, most answers to hazardous operational requests are confident, fluen

Four Tesla V100s from 2017 matched my RTX 5090 on single-request Qwen 3.8 decode. Repo: The 5090 was not being held back. It ran NInfer , a specialist engine built to make this exact model as fast as possible on that GPU. (love this guys work) The V100s ran Qwen3.8's published mixed FP4/FP8 weights unchanged. This should be impossible . NVFP4 was built for Blackwell. The RTX 5090 has native silicon for FP4 and FP8; V100 has none of these advantages. And yet via software I wrote a translator fast

I have been using local models on/off for like 2 years or so but never really used them extensively because the closed ones were always much better. Once Qwen 3.8 27B was released I decided to give it another serious try. I configured Cline and ZooCode as VSCode addons, installed a few MCP servers and added one skill. When I used these tools with Deepseek V4 Flash - they do the job quite well (mostly Home Assistant configuration editing etc.) but it is still way worse than Claude Code/GitHub cop

Byte-level hierarchical language models (LMs) have recently emerged as a robust alternative to their popular counterparts that use subword tokenization. However, generating one byte at a time remains a bottleneck for inference speed. To address this, we introduce multi-byte prediction (MBP), which generates multiple bytes in parallel, speeding up inference with minimal performance impact and no additional parameters. MBP builds on the popular multi-token prediction (MTP) paradigm with two crucia

Self-evolving agents improve future behavior from interaction experience, yet existing evaluations typically optimize under fixed execution conditions and do not test recovery after those conditions change. To address this gap, we introduce PACE-Bench (Physics Adaptation via Code Evolution), a simulator-grounded benchmark of 144 source-to-target adaptation pairs across six physics domains. Each pair links a source environment to a mutated target environment with the same goal and interface. A co

Pixel-space generative models bypass lossy latent compression, yet necessitate joint learning of global structure and fine-grained details in a high-dimensional space. Standard flow matching interpolates noise toward a fixed clean-image endpoint, leaving the spectral evolution to be learned implicitly. In this paper, we introduce Energy-Guided Flow Matching(EG-FM) that explicitly models a coarse-to-fine generative trajectory by moving endpoint. Specifically, EG-FM replaces the fixed endpoint wit

Intermediate tokens, called "thinking" or "reasoning" actually are nothing like it. Humans do step-by-step reasoning leading to the conclusion. LLMs use intermediate traces to augment their prompt . This explains why sometimes the answer is very good but the "reasoning" is verbose. Flooding your context window or fighting compaction are different issues. edit: I love this section from the main research they linked. Our findings consistently challenge the prevailing narrative that intermediate to

Bankrupt Spirit accused of selling out workers in massive data sale to Google.

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. AI’s recursive self-improvement might not come so quickly after all The AI industry’s boldest promise right now is that AI will soon improve itself, with almost no need for human oversight.…

arXiv:2608.17124v1 Announce Type: new Abstract: Combining the answers a large language model (LLM) samples for a question into one decision is a test-time information fusion problem, usually solved by majority voting. Voting is unreliable on difficult questions, where the sampled answers share correlated errors, so the wrong answer can win and drawing more samples makes the decision worse. Selecting a candidate by reading a correctness signal from the model's hidden states is a promising alterna

arXiv:2608.17183v1 Announce Type: new Abstract: Small Language Models (SLMs) are increasingly deployed in resource-constrained, privacy-sensitive settings, where safety and bias failures can cause security and societal risks. However, existing AI safety\slash security\slash compliance benchmarks are designed for large language models that may not transfer reliably to SLMs. We therefore ask: Can these benchmarks effectively and reliably evaluate SLMs? To answer this question, we conduct a large-s

arXiv:2608.17067v1 Announce Type: new Abstract: As text-to-image generative models advance, they raise critical safety concerns, particularly the generation of Not-Safe-For-Work (NSFW) content such as violence and nudity, further exacerbated by red-teaming adversarial attacks. Existing defenses predominantly operate under white-box assumptions, relying on text encoder optimization, weight editing, or inference-time intervention, and fundamentally cannot scale to proprietary models. Black-box alt

Frontier large language models (LLMs) safety evaluation has largely treated harmful generation as an attack outcome rather than as an object of analysis. Consequently, little is known about the harmful outputs produced during model misbehavior, partly because large-scale, high-quality collections of frontier-LLM misbehavior are difficult to obtain. To address this gap, we introduce HarmProfile, a content-centric benchmark dataset that collects model misbehavior across diverse harm categories and

None has undergone post-training, giving researchers flexible starting points for continued pre-training, fine-tuning, and further research. Two key highlights: - They use WSM to replace LR decay with weighted checkpoint merging, making the training process better suited for continual pre-training while enabling offline exploration of different LR decay strategies. - With one shared training recipe, the community can validate strategies on tiny-base, then scale them to flash-base. #1- Ling-3.0-t

The quality and diversity of instruction-based video editing datasets are steadily improving, yet existing datasets mainly focus on single editing operations and fall short in supporting compositional instruction-guided video editing. In particular, multiple editing intents must be jointly understood and faithfully executed within the same video. To address this issue, we introduce CoinVE-200K, a large-scale, high-quality dataset for Compositional Instruction-Guided Video Editing. CoinVE-200K co

Vision encoders are a critical component of vision-language models, and scaling their capacity effectively improves performance. However, dense scaling increases compute cost and inference latency. Mixture-of-Experts (MoE) architectures offer a compelling alternative, having enabled efficient scaling in LLMs, yet the MoE design space for CLIP-style vision encoders remains underexplored at State-of-the-Art (SOTA) levels. In this work, we systematically study MoE designs for vision encoder scaling

Large-scale image generation has benefited from advances in data scale, quality, rebalancing, and recaptioning, yet conventional pipelines typically optimize task-specific datasets in isolation. A central challenge is not only how to curate each task-specific corpus, but also how to organize heterogeneous supervision according to the dependencies among generative capabilities. We present a capability-driven data infrastructure that couples capability-specific supervision construction with capabi

The launch of the new study features marks Google's latest effort to make Gemini the AI assistant that students turn to when learning and studying, as it continues to compete with companies like OpenAI.

Framework says it's replacing some out-of-warranty AMD mainboards.

TerraPower's nuclear power plant possesses a strategic advantage over competitors, especially when chasing after data center deals.

Good day! Over a year ago, I made a post here about Meelo, which had received a lot of positive attention. A few things have changed since, and I thought a lil' update wouldn't hurt :) Meelo is a self-hosted music server, that focuses on UI and metadata integration. It supports duplicates, songs grouping (remixes, instrumentals, etc.), album types (studio, live, compilations, etc.), Music Videos, and other cool stuff. Since my last post (around v3.1.0), new features were added: Meelo now has a c

arXiv:2608.16977v1 Announce Type: new Abstract: AI systems are increasingly capable of contributing to mathematical research. In research practice, frontier-model reasoning is a limited resource, and expert mathematical review is even more sharply constrained. Allocating these scarce resources well is therefore central to making AI-assisted mathematical discovery efficient. In most current AI-for-math workflows, human effort is concentrated at the beginning and end, in selecting suitable researc

High-resolution image editing is increasingly demanded in professional workflows, yet existing diffusion-based models remain constrained to resolutions below 1K due to quadratic attention complexity and prohibitive memory requirements. A prevalent workaround employs a two-stage pipeline: editing at low resolution followed by independent super-resolution. However, this approach suffers from two critical issues: information divergence, where hallucinated details contradict the original high-resolu

arXiv:2608.17170v1 Announce Type: new Abstract: Algorithm selection for constraint satisfaction problems requires extracting features that capture problem structure. Manually designing feature extractors demands deep domain expertise and quickly becomes a bottleneck when new problem classes appear. We present an automated approach that uses Large Language Models (LLMs) in an agentic check--fix--verify loop to synthesize executable Python scripts that act as interpretable, problem-specific featur

arXiv:2608.17150v1 Announce Type: new Abstract: To effectively collaborate with users on knowledge-intensive tasks, Large Language Models (LLMs) must perform information calibration: matching content to a user's evolving understanding and cognitive capacity. Yet user simulators used to evaluate and train LLMs do not explicitly model user knowledge so they neither produce realistic interactions across knowledge levels nor reflect how interactions unfold as that knowledge evolves. To close this ga

AI agents that run in a loop Discussion | Link

Instruction-based general video editing seeks to unify diverse editing operations within a single, intuitive interface. Existing approaches often rely on resource-intensive conditioning, using either heavyweight branches or costly source concatenation. Is there any efficient way to model editing intent? Thus, we introduce GRNEdit, a lightweight two-stage framework. GRN inspires our approach by encoding visual semantics through combinations of bits. Through task-specific fine-tuning, we take this

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. We still don’t know how people are really using AI AI companies like Anthropic and OpenAI regularly publish reports on how people are using their products. But they only release the…

AI workspace for reproducible scientific research Discussion | Link