Search: “reasoning”

Power Limits, Local AI, and Questionable Uses of My Free Time

Good day data nerds! I am trying to get more into running models, learning about agentic workflows, and creating my own tools. But as you do (right?) I had to fine tune my current setup. With the way

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Model Releases

Qwen 3.8 27B vs Qwen 3.8 Flash Next and time to complete a coding task.

DGX agent

I am new to all this so still a lot to learn. If I give Qwen3.8 27B a coding task to fix some bugs in some code, it went through and found and fixed several in like 5 or 10 minutes. I gave qwen flash

model-releasesr-localllama

27 Sept 2026

Local Ai

Qwen3.8-27B FP8 dual GPUs

Hardware RTX 5090 (32 GB) + RTX 4070 Ti Super (16 GB) = 48 GB VRAM 32 GB DDR5 6200 RAM Arch Linux, KDE on the 5090 (takes ~1-1.5 GB VRAM) daily driver OrcaRouter Qwen3.8-27B Uncensored, block-FP8 (28.

local-air-localllama

Qwen3.8-Flash-Next 177B NVFP4(119GiB): SSD streaming at 9-10 tok/s on one 16 GB RTX 5060 Ti + 32 GB RAM

We built an inference engine for MoE models that don't fit in VRAM + RAM. Most of the model stays on the SSD, and experts are read as tokens need them. This started as a proof of concept, and poc work

Splash fork optimised for M5 Max: ~1.5× faster (1.25× single request)

I’ve spent the last couple of days with Opus 5.5 working on a fork of Inco’s excellent and already blazingly fast Splash engine to optimise it for M5 Max chips. Taking liberties and referring to it as

Industry

the inference platforms are doing hardcore work, post-training open weight models to solve customer problems: @FireworksAI_HQ ember-1 @fal H…

the inference platforms are doing hardcore work, post-training open weight models to solve customer problems: @FireworksAI_HQ ember-1 @fal H3 max more to come;) makes total sense considering their pos

industrysonya-huang--x

This is brilliant. Recommended reading. It's a great example of how you could potentially build more effective harnesses using System One an…

This is brilliant. Recommended reading. It's a great example of how you could potentially build more effective harnesses using System One and System Two models. It explores Jev's potential as a fuzzy

agentsdair-ai--x

Zer0Fit - Zero-shot predictions, classifications, and regressions using Google ML research models running locally as a dockerized MCP

AI grad student here. With all the recent interest in Jev, I thought I would share something I built la few months ago that brings ML models and LLMs together in a different way than Jev does for diff

b11195

model-releasesllama-cpp-releases

ggml-cpu: tiled mul_mat for k-quants (#27851) Added tiled mul_mat. For each mul_mat_one_chunk, quants are unpacked into (max) 256x256 tiles of int8, one routine per quent. Then microkernel computes 16

Getting stupidly good results on my 4x3060ti setup.

About a month ago I was having FOMO and was going to spend coin I don't really have on new graphics cards. Instead of doing that though I decided to spend the money on a new motherboard+cpu and try to

How accessible is local AI actually, and what happens if affordable access to frontier models doesn’t last?

Sometimes it’s easy to forget that this sub and others like it are probably the extreme minority when it comes to this hobby. Most people, I would think, don’t use or can’t afford one good GPU, let al

i don’t get it fully but this looks cool

i don’t get it fully but this looks cool We achieved a breakthrough at BTL. AI models are very linear. They think in one direction at a time, and when they're wrong they scramble and start again. For

agentsyohei-nakajima--x

Ling Tiny 3.0 is a glimpse of the future

I've been playing around with Ling 3.0 Tiny, which is an 8 billion parameter model (MoE, 1B active). And I've had a lot of poignant thoughts as a result. Just for fun, I got it running with llama.cpp

This is huge. Xiaomi just open-sourced 7k+ reinforcement learning task environments used to train MiMo, covering code, cybersecurity, genera…

model-releasesclem-delangue--x

This is huge. Xiaomi just open-sourced 7k+ reinforcement learning task environments used to train MiMo, covering code, cybersecurity, general tool use, visual web development, and music-related tasks.

We had reached peak ChatGPT in July

No 5hr limits for Plus, strong Sol in chat that could actually work on problems for a long time. That’s what made me subscribe. Now the product feels weaker somehow, it’s not just the 5hr limits back

model-releasesr-chatgpt

Applications

Worst thing about this is we have no idea if GPT-6 was specifically post trained on Nethack or not. Zero transparency about the training dat…

applicationsethan-mollick--x

Worst thing about this is we have no idea if GPT-6 was specifically post trained on Nethack or not. Zero transparency about the training data and environments. You can't independently evaluate the sig

A Living Benchmark for Information Retrieval from Electronic Health Records

arXiv:2609.30205v1 Announce Type: new Abstract: Large language model (LLM)-based clinical assistants are increasingly being integrated into electronic health record (EHR) systems, transforming how cli

AgileLog: A Forkable Shared Log for Agents on Data Streams

arXiv:2604.14590v3 Announce Type: replace-cross Abstract: In modern data-streaming systems, alongside traditional programs, a new type of entity has emerged that can interact with streaming data: AI a

agentsarxiv-cs-ai

Applications

ARGUS: Role-Aware Event Knowledge Graphs for U.S. Employment-Discrimination Complaints

arXiv:2609.30184v1 Announce Type: new Abstract: U.S. employment-discrimination complaints describe complex event sequences that are not explicitly captured by lexical or embedding-based representation

applicationsarxiv-cs-cl

Banger paper from Microsoft and colleagues. We talk about human taste being important in this AI era. But for recursive self-improvement, an…

Banger paper from Microsoft and colleagues. We talk about human taste being important in this AI era. But for recursive self-improvement, an agent's taste also matters. The big question is: How often

model-releasesdair-ai--x

Banger report on building faster memory for AI Agents. If your agent's memory layer is slow, this design is worth a look. Great weekend read…

Banger report on building faster memory for AI Agents. If your agent's memory layer is slow, this design is worth a look. Great weekend read. It proposes a new agentic memory architecture, Jev-Mem, in

agentsdair-ai--x

BiGraph-Diffuse: A Bidirectional Diffusion Language Model with Graph-Structured Retrieval For Mental Health Counseling

arXiv:2609.29519v1 Announce Type: new Abstract: Mental health disorders affect hundreds of millions of people around the world, yet access to professional counseling remains severely limited. AI-power

C3M: Cross-Session Multimodal Memory Maintenance for Long-Horizon Tasks

arXiv:2609.29735v1 Announce Type: new Abstract: Long-horizon tasks require preserving and later recovering cross-session evidence under a bounded, query-blind memory budget. Existing compression can d

ChunkRank: Model-Aware Text Chunking and Abstention-Aware Answer Selection for LLM Pipelines

model-releasesarxiv-cs-cl

arXiv:2609.29828v1 Announce Type: new Abstract: We present ChunkRank, an open-source Python library that derives chunk boundaries from a target model's tokenizer and context window, and selects an ans

CrossSafe: Towards Cross-Embodiment Latent Safety Filters

arXiv:2609.28984v1 Announce Type: cross Abstract: Cross-embodiment learning has shown that a single model, such as a vision-language-action (VLA) model, can learn state representations and manipulatio

Do World Models Make Better Robots? A Survey of Evaluation Benchmarks for Predictive Embodied Intelligence

arXiv:2609.29669v1 Announce Type: cross Abstract: Robot learning now advances along two tracks that rarely meet. On one side, direct Vision-Language-Action (VLA) policies map observations to actions a

Evaluation of Multi-Turn Consistency in LLM Agents: Survival Analysis and Failure-Rationale Taxonomy

arXiv:2609.29508v1 Announce Type: new Abstract: Large language model (LLM) agents may perform well on isolated tasks yet drift into inconsistency over extended interaction. We evaluate temporal consis

Exploiting Target Knowledge from MLLMs for Robust Few-Shot Segmentation

model-releasesarxiv-cs-cv

arXiv:2609.28949v1 Announce Type: new Abstract: Few-shot segmentation (FSS) aims to segment unseen object categories with a few (e.g., one or five) labeled examples, enabling efficient adaptation to n

Fair Like Us? Auditing LLM Alignment in Resource Allocation

arXiv:2609.29692v1 Announce Type: new Abstract: Fair allocation of scarce, indivisible resources is an important challenge in many societal problems. While there are several formal theories of fairnes

Foundations of Large Language Models

arXiv:2501.09223v3 Announce Type: replace-cross Abstract: This is a book about large language models. As indicated by the title, it primarily focuses on foundational concepts rather than comprehensive

Free the Language Model From the Vision Encoder: Semantic Serialization as a Perception Interface for Small Language Models

model-releasesarxiv-cs-cl

arXiv:2609.29601v1 Announce Type: cross Abstract: End-to-end vision-language models (VLMs) bind visual competence to the scale of their language model: as the language model shrinks, perception and re

From Passive Execution to Active Exploration: Agentic Embodied Manipulation in Realistic Environments

arXiv:2609.29091v1 Announce Type: new Abstract: Recent advances in agentic systems have substantially enhanced the long-horizon capability of embodied manipulation. However, many existing frameworks s

agentsarxiv-cs-ro

GRASP: Generating, Revising, and Assessing for Strategic Planning with Agentic AI

arXiv:2609.30147v1 Announce Type: new Abstract: Large Language Models (LLMs) typically exhibit a performance profile where reliability degrades as task complexity increases. We address the challenge o

Grow the Harness, Not the Context: From Strategy-Free Scaffolds to Reusable Specialist Agents

arXiv:2609.26760v2 Announce Type: replace Abstract: Large language model (LLM) agents often handle streams of related tasks, yet standard harnesses repeatedly ask the model to reconstruct the same con

How long can I expect to wait until the local ~30B A3B frontier catches up to GLM 5.3 Flash quality?

The jump from Qwen3 Coder 30B A3B to current-day Qwen 3.6 35B A3B is crazy, especially with all the fine-tunes, and that was around 6 months (I didn't care for local AI back then, or AI at all, apart

Hardware

I ran the actual break-even math on buying vs renting an H200 box, and it is not where I expected

Every rent-vs-buy thread I read has confident people on both sides, but not many actually show the numbers. So I finally ran the numbers for our own decision. Posting the working here in case it is us

hardwarer-localllama

Instrumental Monitor Evasion Emerges Under Ordinary Task Pressure

arXiv:2609.30217v1 Announce Type: cross Abstract: A central concern in AI safety is that agents may treat oversight as an obstacle when it conflicts with completing their goals. We study instrumental

Jensen Huang tells Ezra Klein that AI is just software whose existential risks are overstated, yet his standards would shut down OpenAI and 10x safety spending (Zvi Mowshowitz/Don't Worry About the Vase)

Zvi Mowshowitz / Don't Worry About the Vase: Jensen Huang tells Ezra Klein that AI is just software whose existential risks are overstated, yet his standards would shut down OpenAI and 10x safety spen

safetytechmeme

Local Ai

Jev-Mobile: Jev as an Executor for Mobile GUI Agents

arXiv:2609.30186v1 Announce Type: new Abstract: Vision-language models (VLMs) have become a common foundation for autonomous mobile GUI agents, but most existing systems rely on the VLM for both plann

local-aiarxiv-cs-ai

Mica v0.1 4B: open Jev-style decision model (yes/no, choice, score) that runs on an 8 GB GPU — trained for under $30 of GPU time

I've been building a small decision model for agent loops: gates, routers, 'should I ask the user or just act' checks. It's out now as Mica v0.1 4B (Apache-2.0). What it does You give it a state, a qu

Local Ai

MILO: Efficient Many-shot In-Context Learning with Block-wise Low-rank Compression

arXiv:2609.29913v1 Announce Type: new Abstract: Many-shot in-context learning (ICL) enables large language models (LLMs) to adapt to complex tasks by conditioning on thousands of demonstration example

local-aiarxiv-cs-cl

Near-Oracle KV Selection via Pre-hoc Sparsity for Long-Context Inference

arXiv:2602.08329v2 Announce Type: replace-cross Abstract: A core bottleneck in large language model (LLM) inference is the cost of attending over the ever-growing key-value (KV) cache. Although near-o

Applications

Neuralized Multi-Wavelet Decomposition for Time Series Classification and Forecasting

arXiv:2609.29317v1 Announce Type: cross Abstract: Time series analysis is fundamental in domains such as finance, healthcare, and meteorology. Real-world time series often exhibit multiscale character

applicationsarxiv-cs-ai

No More Free Lunch: Corpus Task Complexity Matters as Corpora Grow

arXiv:2609.29245v1 Announce Type: cross Abstract: Given a large corpus, the questions one might ask can vary -- from 'When was the first human heart transplant?' to 'What are all the contradictory cla

Tutorials

Nope. It's still true. Where is your domestic robot? Where is your Level-5 self-driving car? Where is your robot car that can learn to drive…

Nope. It's still true. Where is your domestic robot? Where is your Level-5 self-driving car? Where is your robot car that can learn to drive in a few hours of practice like any 17 year old? Where is y

tutorialsyann-lecun--x

Novelty Adaptation Through Hybrid Large Language Model (LLM)-Symbolic Planning and LLM-guided Reinforcement Learning

arXiv:2603.11351v2 Announce Type: replace-cross Abstract: In dynamic open-world environments, autonomous agents often encounter novelties that hinder their ability to find plans to achieve their goals

agentsarxiv-cs-ai

Persuaded, Not Informed: Incentive-Misaligned Witnesses Defeat In-Context Grounding

arXiv:2609.28854v1 Announce Type: cross Abstract: Language-model agents increasingly answer questions over customer-relationship management (CRM) records, such as whether to qualify a sales lead. We i

Post-Training Leaves Behavioral Shadows on Unrelated Decisions

arXiv:2609.29233v1 Announce Type: cross Abstract: We find that language models can transfer capabilities through task-unrelated text. Post-training typically improves language models using task-specif