CompanyAnthropic8 recent entries9 Oct 2026Qwen 3.8 Flash Next-GSQ-RCO-IQ2_XS at ~21 tok/s on just an RTX 3060 12GB + 16GB DDR4 RAM(No gate pruning, 100% bit-exact)(Don't judge by the screenshot, the cache is cold. It hits 24+ tok/s with a warm cache!) About two months ago, I made a post here asking whether predicting which MoE experts would be used on the next →9 Oct 2026I made a portable AI memory format that transfers both context and skillLast month I created a portable memory format as a .txt file as a small project. I named it as: Memory Journal, in a shape of a skill card. It's essentially a large prompt wrapped in a text file. Init
CompanyOpenAI8 recent entries8 Oct 2026Awesome day, @satyanadella! Windows sparked a platform shift that created a new industry for NVIDIA. Then we invented programmable shading G…Awesome day, @satyanadella! Windows sparked a platform shift that created a new industry for NVIDIA. Then we invented programmable shading GPUs for DirectX, which led to CUDA. Then we partnered to bri→8 Oct 2026Anthropic launches OSS Scanner, a free opt-in vulnerability scanner for critical open-source projects; its AI-generated reports are sent without human review (Anthropic)Anthropic: Anthropic launches OSS Scanner, a free opt-in vulnerability scanner for critical open-source projects; its AI-generated reports are sent without human review — We're launching OSS Scanner, →9 Oct 2026Qwen3.8-27B: 159 tok/s on R9700, 64 tok/s on Strix HaloLemonSeed Studio is an iPad editor/IDE with on-device inference on an AMD GPU in a Thunderbolt enclosure. It embeds the unmodified upstream Linux amdgpu + amdkfd driver (mac_linuxgpu) as a PCIDriverKi→9 Oct 2026A new feature for my blog, built using my voiceI shipped a new feature for my blog today: the Newsletters page, which offers an index of all of the newsletters I've sent out, both my free weekly Substack and my monthly sponsors-only updates. I bui→10 Oct 2026Engineer / developer observations of Gemma4-31B, Qwen3.8-27B, and 6.1-Sol for software engineering workFor what it's worth. I'm a mechanical engineer by education and software engineer in practice over the past ~20 years. Playing around with local models has been a recent hobby. Today I decided to do a→10 Oct 2026Benefits of using bigger models than Qwen 3.8 flash next?I recently started using local LLM for mostly coding tasks for my personal projects. I also have found local models responding much better for topics such as health, fitness, ergonomics etc. compared →10 Oct 2026Basalt: Flash-Next at 665 tok/s structured, 354 prose on a 5090 + 5060 Ti (2.6x Strata)Basalt: Blackwell inference engine Hi all! So over the last week I've been working on a Strata fork that's heavily tuned and can achieve throughput up to 2.6x what Strata usually does on the same weig→11 Oct 2026'Strata' for GLM5.3 Flash is here for some! Project MayaI stumbled across this as I was currently having glm5.3 flash run only around 10tok/s basically unusable. With Maya i am running 30 tok/s now. GLM5.3 feels even with this low quant like a much more en
CompanyGoogle8 recent entries7 Oct 2026Google rolls out improved SynthID AI content detector, now available globallyGoogle has launched a global SynthID AI content detector at synthid.com that identifies invisible watermarks embedded in images, videos and audio generated by its Gemini model. The upgraded tool now r→7 Oct 2026Fine-Tuning VLM for Enhancing AI's Spatial Intelligence: Understanding 3D and 2D RotationsarXiv:2610.04206v2 Announce Type: replace Abstract: Spatial intelligence is a fundamental skill in multiple domains, such as Science, Technology, Engineering, and Mathematics (STEM), Medicine, Archite→8 Oct 2026Welcome to Gemini at Work 2026: Introducing the Gemini agentEditor’s note: This article is adapted from Thomas Kurian’s keynote address at Gemini at Work 2026. It includes a number of announcements: Gemini, a single, universal agent for work that answers your →8 Oct 2026GPU - Vulkan llama.cpp benchmarks sorted by price to performanceThis table to help anyone looking to build a budget Data Center homelab. I copied the bulk of value based, mid level, decent speed results GPUs and feed it to AI or SI and here are the recommended res→8 Oct 2026Empowering SMBs to do more with GeminiToday, we announced the new Gemini agent, your single, universal agent for work. For small and midsize businesses, with lean teams and margins, Gemini will be a critical tool for you to operate effici→10 Oct 2026Benefits of using bigger models than Qwen 3.8 flash next?I recently started using local LLM for mostly coding tasks for my personal projects. I also have found local models responding much better for topics such as health, fitness, ergonomics etc. compared →12 Oct 2026MathVis: Can AI Reason through Drawing? A Benchmark for Mathematical Figure GenerationarXiv:2610.12954v1 Announce Type: new Abstract: This paper introduces MathVis, a benchmark designed to evaluate LLMs' ability to visualize frontier mathematical problems and perform diagram-assisted r→12 Oct 2026Introducing DigitalOcean Signals and Simulations: see how your agents and models really behave in production, then test the fix before it shipsYou ship an agent. It mostly works. Then the feedback loop becomes a support ticket: a user got the wrong answer, a tool call looped four times, a refund went out that shouldn’t have. Standard observa
CompanyMeta8 recent entries8 Oct 2026chatgpt's new intelligent ui was reverse engineered in less than 24 hours, and apparently you can recreate it with local llmscame across a pretty interesting technical breakdown of chatgpt's newly launched 'intelligent ui' feature, and thought this subreddit might find it interesting. for anyone unfamiliar with the concept,→8 Oct 2026$2800 rig with 8x Radeon Pro V620 (256 GB VRAM) + custom vLLM fork = Qwen3.8-Flash-Next at 60 to 100 t/s decode and 3000+ t/s prefillPost title is slightly misleading, I don't think you can get these for $350 each anymore but they're still pretty cheap all things considered. They're Radeon Pro V620's which are older RDNA2 enterpris→9 Oct 2026BigMoeOnEdge+ – an Android chat app fork of BigMoeOnEdge (llama.cpp / GGUF)I made BigMoeOnEdge+ (BigMoeOnEdge-plus), a fork of Helldez's BigMoeOnEdge, the engine that streams MoE experts from flash so models bigger than RAM can run on-device. It adds a full Android chat app →10 Oct 2026Qwen Flash Next @ 137 tok/s & 3,497 tok/s Prefill w/ 512k context on a 5090, 192gb ram, Windows Build, comparing Strata and InfernixStrata vs Infernix on the same model, RTX 5090 — A/B at 262k and 512k context Hi folks, another inference engine post for your feed. I was blown away by Strata, but saw someone else post about Inferni→10 Oct 2026Improve token per second without touching quantSpent the past month tweaking and experimenting with many different numbers to achieve 30tps. Hardware: -Rx6700xt 12gb vram (AMD) -2x16 ddr4 3200 ram -r5 5600x -llama.cpp vulkan sdk -window11 (no wsl →10 Oct 2026Created laya : Now Introducing a new 800 Million Param physics-based typed decision model with 73k context and image supportAround 3 weeks ago, I posted on this subreddit about how Jev used my one-year-old architecture, og post: https://www.reddit.com/r/LocalLLaMA/s/VpuMJt577S Then, just after that, I introduced Laya, the →12 Oct 2026Taking Tokenization off the Host for Agentic LLM ServingarXiv:2610.13325v1 Announce Type: cross Abstract: Agentic workloads resend long contexts that barely change between calls. Prefix caching now skips most of their prefill, yet every request must still →12 Oct 2026Meta-Reflective Evolution Effectively Improves Production AI SystemsarXiv:2610.12794v1 Announce Type: cross Abstract: Large Language Model (LLM)-guided evolutionary optimization has successfully improved prompts, programs, and even entire AI systems. Most optimization
CompanyMistral8 recent entries29 Sept 2026Accelerating agentic RL and evaluation research velocity with 45x faster GKE Agent SandboxWhen scaling up agentic reinforcement learning (RL) and evaluation across massive parallel rollouts, frontier AI labs inevitably hit a bottleneck: Expensive GPU clusters sit idle, waiting minutes for →30 Sept 2026Follow-up: my native Rust + Vulkan Transformer training backend — 14 days later, now 14 parity-verified architectures and full PEFTFollow-up to my post from about two weeks ago. A lot has changed since then, so I wanted to post an update on where the backend is now. Where the green architectures stand When I posted last time, 7 a→6 Oct 2026Mistral Large 4 is one of the world's strongest AI models for cybersecurity.Mistral AI announced that its new model, Mistral Large 4, is among the world's most powerful AI systems specifically designed for cybersecurity tasks. The tweet shared on October 6, 2026 highlights th→6 Oct 2026Mistral Large 4 against the field: efficiently reasoning over diverse complex challenges. CTF speedrun: 19 challenges, with tool calls and s…Mistral Large 4 against the field: efficiently reasoning over diverse complex challenges. CTF speedrun: 19 challenges, with tool calls and solve times grounded in actual runs. It solved 18 of the 19. →6 Oct 2026Malware reverse-engineering: solving an out-of-distribution investigation task. When faced with an unknown binary, Mistral Large 4 reverse-e…Malware reverse-engineering: solving an out-of-distribution investigation task. When faced with an unknown binary, Mistral Large 4 reverse-engineers it end-to-end. In this case, it concludes the sampl→6 Oct 2026For their Mistral Large 4 release, @MistralAI ran a blind Surge human eval, with our expert software engineers evaluating five frontier mode…For their Mistral Large 4 release, @MistralAI ran a blind Surge human eval, with our expert software engineers evaluating five frontier models on coding quality. Le Chonk ranked #1 among open-weight m→8 Oct 2026We serve those as well! Sorry if using nuclear energy is a real blockerWe serve those as well! Sorry if using nuclear energy is a real blocker Berlin-based search engine Ecosia, used by government services, is dropping its French AI partner, Mistral, its CEO told POLITIC→9 Oct 2026I’m in Seoul building Mistral’s founding Korea team. Looking for kind, ambitious researchers and engineers to build and customise LLMs for r…I’m in Seoul building Mistral’s founding Korea team. Looking for kind, ambitious researchers and engineers to build and customise LLMs for real problems, from data and evals to RL and training at scal
CompanyxAI8 recent entries30 Sept 2026SpaceXAI just literally shared its engineering team’s Grok Bots with everyone And Grok Bots just got much more powerful for building softwar…SpaceXAI just literally shared its engineering team’s Grok Bots with everyone And Grok Bots just got much more powerful for building software Bots can now hand off coding tasks directly to Cursor, man→1 Oct 2026VAmoS Part Deux: Harder, More Realistic Voice-Agent SimulationarXiv:2609.38512v1 Announce Type: new Abstract: Voice agents in production must handle several requests, background speech, and customers who lose patience. We introduce VAmoS Energy, a benchmark that→4 Oct 2026Synthesis: Elon-Musk--XAuto-generated synthesis of 2431 entries about elon-musk--x→4 Oct 2026SpaceXAI engineer, Lauren Tan: 'I was a meat proxy between my agent and my browser. So I fired myself Now 10+ Chiefs of Staff run my agents …SpaceXAI engineer, Lauren Tan: 'I was a meat proxy between my agent and my browser. So I fired myself Now 10+ Chiefs of Staff run my agents 24/7 and I just check the work in the morning' In 1 hour, sh→4 Oct 2026Wiki Lint Report — 2026-10-04Automated lint: 19 errors, 64915 warnings, 3 info→6 Oct 2026Do RUL explanations hold up? Faithfulness and stability of attributions on C-MAPSSarXiv:2610.04278v1 Announce Type: new Abstract: Deep remaining-useful-life (RUL) models on NASA C-MAPSS are now routine, and so are heatmaps that colour sensors and timesteps. A heatmap that looks mec→7 Oct 2026Introducing MCP tools for @dair_ai. Connect the MCP to your Codex/Claude/Grok Bot to discover and explore the top AI papers. We carefully cu…Introducing MCP tools for @dair_ai. Connect the MCP to your Codex/Claude/Grok Bot to discover and explore the top AI papers. We carefully curate the papers, so you can expect mostly bangers. The index→7 Oct 2026After many years in top machine learning organizations, with friends at frontier labs, I still consider Tesla’s ML engineers the best—the on…After many years in top machine learning organizations, with friends at frontier labs, I still consider Tesla’s ML engineers the best—the only ones I would trust with my life. Unsatisfied with theory,
CompanyDeepSeek8 recent entries8 Oct 2026Open sourcing Game Summoner, my prompt to game sitetldr: Open sourced my prompt to game suite: https://github.com/ndamiano/ai-agent-test, it's MIT licensed, and this runs well on my 5090 with 64gb ram, but any model that can handle tool calls can mana→8 Oct 2026$2800 rig with 8x Radeon Pro V620 (256 GB VRAM) + custom vLLM fork = Qwen3.8-Flash-Next at 60 to 100 t/s decode and 3000+ t/s prefillPost title is slightly misleading, I don't think you can get these for $350 each anymore but they're still pretty cheap all things considered. They're Radeon Pro V620's which are older RDNA2 enterpris→9 Oct 2026Just another purely open-weight models benchmarkLive results, coding benchmarks included, agentic benchmarks included, domains and other classifications filters included. (Spoiler: DeepSeek-V4-Vision-Exp rules, but other models have their rule area→9 Oct 2026I made a portable AI memory format that transfers both context and skillLast month I created a portable memory format as a .txt file as a small project. I named it as: Memory Journal, in a shape of a skill card. It's essentially a large prompt wrapped in a text file. Init→9 Oct 2026A Closer Look at Agentic BBO: Benchmarking LLM Agents for Black-Box OptimizationarXiv:2610.12183v1 Announce Type: cross Abstract: Black-box optimization (BBO) arises in many scientific and engineering problems where objective evaluations are expensive and limited. Recent large la→10 Oct 2026Qwen Flash Next @ 137 tok/s & 3,497 tok/s Prefill w/ 512k context on a 5090, 192gb ram, Windows Build, comparing Strata and InfernixStrata vs Infernix on the same model, RTX 5090 — A/B at 262k and 512k context Hi folks, another inference engine post for your feed. I was blown away by Strata, but saw someone else post about Inferni→11 Oct 2026Qwen3.8 Flash Next fixed my GNOME extensionI love Dash2Dock Lite, but Icedman is always a week or two before updates. I randomly ran pacman for the first time in weeks, not realizing GNOME had a major update. I told it to fix it in DeepSeek Ha→12 Oct 2026Taking Tokenization off the Host for Agentic LLM ServingarXiv:2610.13325v1 Announce Type: cross Abstract: Agentic workloads resend long contexts that barely change between calls. Prefix caching now skips most of their prefill, yet every request must still
CompanyNVIDIA8 recent entries8 Oct 2026GPU - Vulkan llama.cpp benchmarks sorted by price to performanceThis table to help anyone looking to build a budget Data Center homelab. I copied the bulk of value based, mid level, decent speed results GPUs and feed it to AI or SI and here are the recommended res→8 Oct 2026Awesome day, @satyanadella! Windows sparked a platform shift that created a new industry for NVIDIA. Then we invented programmable shading G…Awesome day, @satyanadella! Windows sparked a platform shift that created a new industry for NVIDIA. Then we invented programmable shading GPUs for DirectX, which led to CUDA. Then we partnered to bri→9 Oct 2026Just another purely open-weight models benchmarkLive results, coding benchmarks included, agentic benchmarks included, domains and other classifications filters included. (Spoiler: DeepSeek-V4-Vision-Exp rules, but other models have their rule area→10 Oct 2026Strata with Qwen3.8 Flash Next UD-Q4_K_XLBeen seeing quite a few posts about Strata lately, so I figured I'd give it a shot on my RTX PRO 6000 and I am very impressed. Most of the benchmarks I've seen are using IQ2 or IQ3 quants, so I wanted→10 Oct 2026Basalt: Flash-Next at 665 tok/s structured, 354 prose on a 5090 + 5060 Ti (2.6x Strata)Basalt: Blackwell inference engine Hi all! So over the last week I've been working on a Strata fork that's heavily tuned and can achieve throughput up to 2.6x what Strata usually does on the same weig→11 Oct 2026'Strata' for GLM5.3 Flash is here for some! Project MayaI stumbled across this as I was currently having glm5.3 flash run only around 10tok/s basically unusable. With Maya i am running 30 tok/s now. GLM5.3 feels even with this low quant like a much more en→12 Oct 2026Taking Tokenization off the Host for Agentic LLM ServingarXiv:2610.13325v1 Announce Type: cross Abstract: Agentic workloads resend long contexts that barely change between calls. Prefix caching now skips most of their prefill, yet every request must still →12 Oct 2026A Record of the Process of Increasing MFU from 21% to 50% While Pre-training 21B MoE Using a Single GH200 Chip (Includes Code and Logs)I am an independent researcher. Due to budget constraints, I shifted my 21B MoE pre-training—which I had originally planned to run on 64 H100s—to a single NVIDIA GH200 ($2.29 per hour), and I have com