Model Architecture

Transformers, mixture of experts, attention mechanisms, model design

746 articles across 192 editions

Articles

  1. [Editorial] Cirrius Tech: High confidence is not a security control -- 2026-10-05
  2. [Editorial] Cotool.ai -- 2026-10-05
  3. [Editorial] System 0: Transforming Artificial Intelligence into a Cognitive Extension -- 2026-09-30
  4. Three Days in August: What a DDoS Attack Exposed in Our Network -- 2026-09-30
  5. [Editorial] Auditors find EU cyberattack response weakened by overlapping systems and secretive member states -- 2026-09-30
  6. [Editorial] Pentagon: demand for cyber operations exceeds supply -- 2026-09-30
  7. Trained locally: ultra-fast 0.8B/2B System 1 decision models that match Jev on benchmarks and Doom, ~30 ms per decision (open weights) -- 2026-09-30
  8. Jeeves. Reasoning improves Jev-like decision models -- 2026-09-30
  9. BAAI/AREX-2 - 27B - Agent model based on Qwen3.8 27B -- 2026-09-30
  10. 400+ LLM agents living in a 2004-era MMO server, all local on Qwen3-4B -- 2026-09-30
  11. Splash fork optimised for M5 Max: ~1.5× faster (1.25× single request) -- 2026-09-30
  12. Accelerating vision-language models with LFM2.5-VL-DSpark -- 2026-09-30
  13. [NeurIPS 2026] Concurrent Image Understanding and Generation:Self-Correcting Coupled Markov Jump Processes -- 2026-09-30
  14. Qwen-family LLMs are quietly becoming the backbone of modern audio models; One chart for the architectures of 100+ audio models -- 2026-09-30
  15. lodestones/Kroma -- 2026-09-30
  16. bosonai/higgs-audio-v3-tts-4b -- 2026-09-30
  17. Jev is the FIRST of a Whole New Class of AI Models (Here's How to Actually Use It) -- 2026-09-29
  18. Run decision models on vLLM and Red Hat AI using DiffusionGemma -- 2026-09-29
  19. RedHatAI/diffusiongemma-26B-A4B-it-FP8-dynamic on Hugging Face -- 2026-09-29
  20. FIDES: A Concordance Protocol for LLM-Generated Trading Strategies -- 2026-09-25
  21. [Editorial] Contrastive-LM/CLM: contrastive language modeling -- 2026-09-25
  22. Opus 5.5 is the real deal: same accuracy as Fable 5.1 on 25 graded tasks, faster and way cheaper -- 2026-09-25
  23. Cactus Needle 3: A Sliceable 8-29MB Automation Foundation Model That Matches DeepSeek v4 Flash -- 2026-09-18
  24. [Editorial] GLiNER2.5: span-free information extraction -- 2026-09-18
  25. Made the horizontal open-source model for Jev with RLCD, and it surpasses all the Jev benchmarks. HF space, benchmark, model, repo -- 2026-09-18
  26. XingChen-AGI/Xing4.0-29B-A4B MoE -- 2026-09-17
  27. Qwen3.5 4B + grabbing logits is almost "Jev"? Or even just Qwen Reranker? -- 2026-09-17
  28. oboroge0/hayamimi -- 2026-09-17
  29. Show HN: Kinesis – Control your Mac with the Meta Neural Band -- 2026-09-17
  30. amap-cvlab/ABot-Recon -- 2026-09-15
  31. OUI-1: a model that generates bespoke UI elements -- 2026-09-15
  32. Comfy-Org/Ideogram-4 -- 2026-09-15
  33. ScottStevenson/SuperAstra -- 2026-09-15
  34. A super fast, non-expensive alternative to motion capture - [ft. Sara Silkin] -- 2026-09-15
  35. 1-bit 27B in the browser: 25–30 tok/s on a 6 GB RTX 3060 Laptop (WebGPU, no install) -- 2026-09-15
  36. CPU Only Experimental Sloppy Deepseek V4.1 Flash -- 2026-09-15
  37. 2×RTX 3090 + EPYC box running qwen3.8-flash-next at ~38 tok/s -- 2026-09-15
  38. Koboldcpp v1.121 released -- 2026-09-15
  39. MTP released for Qwen3.8-Flash-Next-GGUF -- 2026-09-04
  40. I pushed Qwen3.8-27B to 2.000 prefill per second and 132 decode per second on A RTX 3090. -- 2026-09-04
  41. Qwen 3.8 27B NVFP4, in a single 5090, using nInfer above 200tps at 180K contexts -- 2026-09-04
  42. R9V: A designer set of kernels I've been working on for R9700s/RDNA4. Qwen3.8-Flash-Next Unsloth IQ4_XS (w/ TP on 2 R9700s, MTP, SSD n-gram, 128k ctx, vision): TG256 of *78 tok/s* (~3x increase), PP8192 of *1510 tok/s* (~30x increase). -- 2026-09-04
  43. GLM-5.3 weights will be released tomorrow -- 2026-09-01
  44. GLM 5.3 and GLM 5.3 Flash ran locally on RTX PRO 6000 WS and built a penthouse using BlenderMCP -- 2026-09-01
  45. friendly reminder you can legally torrent ai models. -- 2026-09-01
  46. [Editorial] -- 2026-08-27
  47. The Hugging Face incident and the road ahead -- 2026-08-27
  48. How Complex Systems Fail (1998) -- 2026-08-27
  49. [Editorial] Researchers Complain That OpenAI Revoked Their Access to Limited Cyber Program -- 2026-08-24
  50. [Editorial] Smolbox — Live Site -- 2026-08-24
  51. [Editorial] Smolbox — remyhax Writeup -- 2026-08-24
  52. [Editorial] gadievron/raptor — Issue #889 -- 2026-08-24
  53. [Editorial] Editor-Curated Video Pick #2 -- 2026-08-24
  54. [Editorial] Editor-Curated Video Pick #1 -- 2026-08-24
  55. [Editorial] sw30labs/singularity-atlas -- 2026-08-24
  56. [Editorial] Ox Alpha: Stealth AI Model With 1M-Token Context -- 2026-08-24
  57. [Editorial] Coding Model Ox Alpha Retains Every Prompt — And You Can't Name the Company Holding Them -- 2026-08-24
  58. Ox Alpha stealth model: GLM5 Air, Mimo V3 or ? -- 2026-08-24
  59. Artificial Analysis' Qwen3.8-27B benchmarks put it neck and neck with DeepSeek V4 and GPT-5.6 Luna Max -- 2026-08-21
  60. [Editorial] YouTube feature -- 2026-08-21
  61. DeepSeek-V4-Flash-Vision-Exp -- 2026-08-21
  62. Microsoft ran a 100B BitNet test on one CPU. Here is the detail the headline misses. -- 2026-08-21
  63. A Preview of DuckDB v2.0 -- 2026-08-18
  64. Reticulum – Decentralized Mesh Network -- 2026-08-18
  65. Design 3D-printable parts by talking -- 2026-08-18
  66. Store Tunes on Paper and Stream Them Over LoRA -- 2026-08-18
  67. [Editorial] GLM-5.3 Release (z.ai) -- 2026-08-14
  68. [Editorial] LinkedIn Post (gMy2vbnc) -- 2026-08-14
  69. Mojo 1.0 -- 2026-08-13
  70. DeepSeek Harness -- 2026-08-13
  71. OpenAI on upcoming model "Astra" (GPT-6): "We're treating it as our first "critical" model for cybersecurity" -- 2026-08-12
  72. Everything you do is being recorded -- 2026-08-12
  73. Underestimated budget solution: radeon 780m iGPU -- 2026-08-10
  74. Building a zero-dependency C inference engine for BitNet (1.58-bit) - lessons from hitting 36 tok/s on a Xeon CPU -- 2026-08-10
  75. I got tired of my 300GB model loads taking 5min on RPC. PR 26291 speeds it 300% to 1min30sec (4060ti+ddr4) + (4060ti+ddr5) -- 2026-08-10
  76. Early signs that Muse-Glimmer-30B might quantize *very* well? Share your experiences. -- 2026-08-10
  77. Best Local LLMs - August 2026 -- 2026-08-10
  78. Qwen3-TTS voice cloning is now in mainline llama.cpp — the old demo finally became real support -- 2026-08-10
  79. Maple-Preview — Ternary 20B MoE running at 120 tok/s on an iPhone -- 2026-08-05
  80. Qwen Developers' AMA: 3.8-27B coming soon, 2.4T params (95B active) -- 2026-08-05
  81. AI9Stars released G9v3-39A5B — Apache 2.0, 39B/5B active MoE -- 2026-08-05
  82. [Editorial] Unsloth Kimi K3 Model Documentation -- 2026-08-05
  83. [Editorial] Series of Model Tests and Results -- 2026-08-05
  84. SciCodePile: A 128GB Corpus and Executable Benchmark for Scientific Code Generation -- 2026-07-31
  85. Procedural desert explorer built with Claude Code (Opus 5) and Three.js -- 2026-07-31
  86. img2threejs — Image-to-3D procedural Three.js model generation -- 2026-07-31
  87. Kimi K3 Architecture Overview and Notes -- 2026-07-31
  88. DeepSeek-V4-Flash-0731 on HuggingFace -- 2026-07-31
  89. [Editorial] Poolside Laguna-S-2.1 — Open Coding Model -- 2026-07-31
  90. LFM2.5-Encoders: Fast at Long Context, Even on CPU -- 2026-07-31
  91. Dario Amodei: closed-weights models are worse than open-weights ones? -- 2026-07-31
  92. [Editorial] -- 2026-07-30
  93. [Editorial] -- 2026-07-30
  94. [Editorial] -- 2026-07-30
  95. Was waiting for Kimi 3 and now Ollama release it and has to pay extra to use it (like OpenRoute) -- 2026-07-30
  96. Everyone posts day-one impressions. What's still in your stack a month later? -- 2026-07-30
  97. [Editorial] -- 2026-07-30
  98. Show HN: Open-source engine running Gemma 4 26B in 2 GB RAM on any M-series Mac -- 2026-07-30
  99. STEEL: Sparsity-Aware Fused Attention for Energy-Efficient Long-Sequence Inference on AMD's XDNA NPU -- 2026-07-30
  100. [Editorial] -- 2026-07-27
  101. [Editorial] -- 2026-07-27
  102. Model Routing Is Simple. Until It Isn't. — IBM Research -- 2026-07-20
  103. [Editorial] OmniRoute — Model Routing Framework -- 2026-07-20
  104. I Burned All My Tokens Researching How to Save Tokens -- 2026-07-20
  105. [Editorial] GCF-Rust — Blackwell Systems GPU Compute Framework -- 2026-07-20
  106. Running Gemma 4 26B at 5 tokens/sec on a 13-year-old Xeon with no GPU -- 2026-07-17
  107. LM Studio Bionic: the AI agent for open models -- 2026-07-17
  108. Prism-ML Bonsai Qwen 3.6 27B -- 2026-07-17
  109. i tried ternary decomposition instead of quantization. it works as good at q4km but takes slightly more vram. while being completly ternary. and completly PTQ (no QAT) -- 2026-07-17
  110. Prayers requested for an Abomination(Modified Gemma) -- 2026-07-17
  111. Hermes on Android (Graphene OS) -- 2026-07-17
  112. nvidia/Nemotron-Labs-Audex-30B-A3B · Hugging Face -- 2026-07-08
  113. Gemma 4 Technical Report -- 2026-07-08
  114. Hugging Face and Cerebras bring Gemma 4 to real-time voice AI -- 2026-07-08
  115. Cloudflare Meerkat - Globally distributed consensus -- 2026-07-08
  116. Run AI workloads on any cloud, store on Hugging Face: zero-egress storage with SkyPilot -- 2026-07-08
  117. [Editorial] Video Submission -- 2026-07-01
  118. Accio-Lab/Dressage -- 2026-07-01
  119. dondai1234/agent-browser -- 2026-07-01
  120. Emry: an event-sourced, local-first observability engine for long training runs -- 2026-07-01
  121. LongCat-2.0, a large-scale MoE model with 1.6T total and 48B Active -- 2026-07-01
  122. [Editorial] Google Gemini Model Family Updates -- 2026-07-01
  123. InternScience/Agents-A1 · Hugging Face -- 2026-07-01
  124. [Editorial] Anthropic Redeploying Fable 5 -- 2026-07-01
  125. [Editorial] Anthropic / White House Lifting AI Export Limits -- 2026-07-01
  126. Why Dario is on fire: lesson from dotcom bubble. -- 2026-07-01
  127. youtube.com -- 2026-07-01
  128. [Editorial] -- 2026-06-27
  129. [Editorial] -- 2026-06-27
  130. trotsky1997/OpenFugu -- 2026-06-27
  131. SPIRAL: Learning to Search and Aggregate -- 2026-06-27
  132. [Editorial] -- 2026-06-27
  133. US government allows Anthropic limited release of AI model that sparked cybersecurity concerns | CNN Business -- 2026-06-27
  134. Fable 5 return RUMORED with some hints in CC -- 2026-06-27
  135. [Editorial] -- 2026-06-27
  136. Pretraining Recurrent Networks without Recurrence -- 2026-06-26
  137. Towards Direct Latent-Space Synthesis for Parallel Branches in LLM-Agent Workflows -- 2026-06-26
  138. KaLM-Reranker-V1: Fast but Not Late Interaction for Compressed Document Reranking -- 2026-06-26
  139. [Editorial] arXiv 2606.24597 -- 2026-06-26
  140. GLM-5.2 is the first open-weights model to cross 80% on Terminal-Bench -- 2026-06-19
  141. unsloth GLM-5.2-GGUF, including 2bit at 238GB -- 2026-06-19
  142. Running local models is good now -- 2026-06-19
  143. Gemma 12b less than 10 watts 6.5pp 1.3tg -- 2026-06-19
  144. Rio 3.5 397B could've simply been a semi-failed embezzling of funding -- 2026-06-19
  145. Concept frustration: Aligning human concepts and machine representations -- 2026-06-18
  146. [Emerging Ideas] Artificial Tripartite Intelligence: A Bio-Inspired, Sensor-First Architecture for Physical AI -- 2026-06-18
  147. Probability-Conserving Flow Guidance -- 2026-06-18
  148. VibeThinker-3B: what is this witchcraft? Killing it at MathQA like it has ~30B parameters -- 2026-06-18
  149. Get in here: Community model build thread -- 2026-06-18
  150. bartowski/command-a-plus-05-2026-GGUF · Hugging Face -- 2026-06-18
  151. nvidia/Nemotron-Labs-Diffusion-14B -- 2026-06-18
  152. Software is made between commits -- 2026-06-12
  153. The Road to the WASM Component Model 1.0 -- 2026-06-12
  154. macOS 27 Beta breaks the ability to boot Asahi Linux -- 2026-06-12
  155. Deficient executive control in transformer attention -- 2026-06-11
  156. Generalization in LLM Problem Solving: The Case of the Shortest Path -- 2026-06-11
  157. zengxiao-he/tessera -- 2026-06-11
  158. Tencent-Hunyuan/UniRL -- 2026-06-11
  159. SafeAdapt: Provably Safe Policy Updates in Deep Reinforcement Learning -- 2026-06-11
  160. Physics-Grounded Multi-Agent Architecture for Traceable, Risk-Aware Human-AI Decision Support in Manufacturing -- 2026-06-11
  161. When Graph Structure Becomes a Liability: A Critical Re-Evaluation of Graph Neural Networks for Bitcoin Fraud Detection under Temporal Distribution Shift -- 2026-06-02
  162. Introducing Mellum2: A 12B Mixture-of-Experts Model by JetBrains -- 2026-06-02
  163. Tencent Hy-MT2 is now under Apache License 2.0 -- 2026-06-02
  164. fxyz666/LogicPipe -- 2026-06-02
  165. [OSS] dlmserve - first serving engine for diffusion language models -- 2026-06-02
  166. veryyoldman/Genspark-AI -- 2026-06-02
  167. Qwen/Qwen-Image-Bench · Hugging Face -- 2026-06-02
  168. Open-source LLMs administer maximum electric shocks in a Milgram-like obedience experiment -- 2026-05-27
  169. [Editorial] Editor's Pick (Video) -- 2026-05-27
  170. Anthropic researcher: "We keep finding things [inside AI models] that are unsettling" — evidence of introspection mirroring joy, fear, grief -- 2026-05-27
  171. lordx64/Qwen3.6-35B-A3B-Claude-4.7-Opus-Reasoning-Distilled -- 2026-05-22
  172. Jackrong/Qwopus3.5-9B-Coder-GGUF · Hugging Face -- 2026-05-22
  173. Qwen 3.6 35B GGUF: NTP vs MTP quantization results across GPUs and CPUs -- 2026-05-22
  174. [WIP] Gemma 4 MTP -- 2026-05-22
  175. Lemonade v10.5.1: an MTP + ROCm 7.13 quick start for Strix Halo -- 2026-05-22
  176. gemma-4-Ortenzya-The-Creative-Wordsmith-31B-it-uncensored-heretic is Out Now -- 2026-05-22
  177. Gemma-4-Gembrain-31B-it-uncensored-heretic Is Out Now -- 2026-05-22
  178. Newbie vibe coding experience: Shifting from Claude Sonnet 4.6 to Qwen3.6-35B-A3B-UD-Q6_K -- 2026-05-22
  179. HalBench: Open Sycophancy and Hallucination Benchmark — 12,800 Graded Responses -- 2026-05-21
  180. DystopiaBench: 42 LLMs tested on willingness to build dystopian systems -- 2026-05-21
  181. Guardrails take an 8B model from 53% to 99% on agentic tasks [ACM CAIS '26 preprint] -- 2026-05-21
  182. bytedance released an open source model that attempts to do just about anything with only 3b parameters -- 2026-05-19
  183. Sapient Intelligence releases HRM-Text 1B: 40B tokens, ~$1k pretrain, beats Llama3.2 3B on MATH and DROP -- 2026-05-19
  184. PaddleOCR 3.5: Running OCR and Document Parsing Tasks with a Transformers Backend -- 2026-05-19
  185. [Editorial] -- 2026-05-18
  186. [Editorial] -- 2026-05-18
  187. [Editorial] -- 2026-05-18
  188. [Editorial] -- 2026-05-18
  189. EMO: Pretraining mixture of experts for emergent modularity -- 2026-05-13
  190. [Editorial] -- 2026-05-13
  191. iai-mcp: Persistent Claude Memory Daemon — 5 Months of Daily Use, Now Open Source -- 2026-05-08
  192. [Editorial] Claude Skins — Custom Claude Code Personas -- 2026-05-08
  193. [Editorial] AI Agents Projects & Tutorials Collection -- 2026-05-08
  194. "Second Thoughts" — A small transformer reads output and feeds it back as a refinement loop, drastically improving a 1.7B model's coding -- 2026-05-06
  195. A plug-n-play open-source pruning tool that is workload-aware (Sculpt) -- 2026-05-06
  196. "I" is not singular — 4 LLM agents with per-agent LoRA on a single RTX 3070 8GB -- 2026-05-06
  197. I made a visualizer for Hugging Face models (hfviewer.com) -- 2026-05-06
  198. The Road to a Billion-Token Context -- 2026-05-06
  199. How OpenAI delivers low-latency voice AI at scale -- 2026-05-06
  200. ArmSSL: Adversarial Robust Black-Box Watermarking for Self-Supervised Learning Pre-trained Encoders -- 2026-04-28
  201. GTFOBins -- 2026-04-28
  202. Anthropic admits to have made hosted models more stupid, proving the importance of open weight, local models -- 2026-04-24
  203. Switching from Opus 4.7 to Qwen-35B-A3B -- 2026-04-24
  204. Forgive my ignorance but how is a 27B model better than 397B? -- 2026-04-24
  205. NSA is using Anthropic's Mythos despite blacklist -- 2026-04-22
  206. [Editorial] Mad Bugs: Reverse Engineering Deep Dive -- 2026-04-22
  207. Anyone deployed Kimi K2.6 on their local hardware? -- 2026-04-21
  208. 24/7 Headless AI Server on Xiaomi 12 Pro (Guide & Benchmarks) Gemma4 VS Qwen2.5 -- 2026-04-21
  209. Gemm4:e4B-IT good at instructions following no refusals. -- 2026-04-21
  210. [Editorial] arxiv:2506.02153 — AI Research -- 2026-04-20
  211. [Editorial] arxiv:1503.02531 — Classic ML/AI Paper -- 2026-04-20
  212. [Editorial] arxiv:2604.06169 — Recent AI Research -- 2026-04-20
  213. [Editorial] Why Your LLM Is Slow (and the Eight Fixes) -- 2026-04-20
  214. Hot Experts in your VRAM! Dynamic expert cache in llama.cpp for 27% faster token generation -- 2026-04-20
  215. [Editorial] Schmidhuber: The World Model Boom -- 2026-04-20
  216. [Editorial] Schmidhuber: World Models, Planning & Curiosity (1990 Origins) -- 2026-04-20
  217. [Editorial] The Ontology Problem (Technical) — Kurt Cagle -- 2026-04-20
  218. Gemma 4 26B fabricated an entire code audit. I have the forensic evidence from the database. -- 2026-04-16
  219. [Editorial] Looped LLMs Are the Nuclear Fusion of AI -- 2026-04-16
  220. Supply-Chain Poisoning Attacks Against LLM Coding Agent Skill Ecosystems -- 2026-04-14
  221. elastic/supply-chain-monitor -- 2026-04-14
  222. [Editorial] UK AI Security Institute + Claude -- 2026-04-14
  223. [Editorial] -- 2026-04-13
  224. [Editorial] -- 2026-04-13
  225. Muse Spark: Scaling towards personal superintelligence -- 2026-04-13
  226. [Editorial] -- 2026-04-13
  227. [Editorial] -- 2026-04-13
  228. Abliterating Qwen3.5-397B on a Mac Studio revealed that MoE models encode refusal differently than dense models — safety refusals route through expert selection and survive weight-baking -- 2026-04-09
  229. Finally Abliterated Sarvam 30B and 105B! -- 2026-04-09
  230. Qwen3.6-Plus -- 2026-04-07
  231. Omnivoice - 600+ Language Open-Source TTS with Voice Cloning and Design -- 2026-04-07
  232. [Editorial] Generative Models and High-Quality Output -- 2026-04-07
  233. Falcon Perception -- 2026-04-02
  234. TRL v1.0: Post-Training Library Built to Move with the Field -- 2026-04-02
  235. lucas-maes/le-wm -- 2026-04-02
  236. Toward explaining why traditional ablation/abliteration works -- 2026-04-02
  237. [Editorial] ramimac/teampcp -- 2026-03-27
  238. This Week in Security: Second Verse, Worse Than the First -- 2026-03-27
  239. Towards a Neural Debugger for Python -- 2026-03-26
  240. Mathematics behind extreme quantization of Microsoft's BitNet -- 2026-03-26
  241. A.T.L.A.S - Adaptive Test-time Learning and Autonomous Specialization -- 2026-03-26
  242. Show HN: Three new Kitten TTS models – smallest less than 25MB -- 2026-03-23
  243. Flash-MoE: Running a 397B Parameter Model on a Laptop -- 2026-03-23
  244. [Editorial] Distillation Techniques -- 2026-03-23
  245. Jackrong/Qwen3.5-2B-Claude-4.6-Opus-Reasoning-Distilled-GGUF -- 2026-03-23
  246. Local Qwen 8B + 4B completes browser automation by replanning one step at a time -- 2026-03-23
  247. Does imatrix calibration data affect writing style? I ran a blind-scored experiment -- 2026-03-23
  248. Memory Chip Crunch to Persist Until 2030, SK Hynix Chairman Says -- 2026-03-19
  249. Meta's renewed commitment to jemalloc -- 2026-03-19
  250. We all had p2p wrong with vllm so I rtfm -- 2026-03-19
  251. [Editorial] LLM Architecture Gallery -- 2026-03-17
  252. Mistral small 4 PR on transformers. -- 2026-03-17
  253. I spent a weekend doing layer surgery on 6 different model architectures. There's a "danger zone" at ~50-56% depth. -- 2026-03-17
  254. Nemotron 3 Super and the no free lunch problem -- 2026-03-17
  255. Timber: AOT Compiler Turns XGBoost/sklearn/ONNX into Native C99 — 336x Faster -- 2026-03-13
  256. [Editorial] Living on the Edge: Advancements in Model Deployment -- 2026-03-13
  257. [Editorial] Translating a PDF While Preserving Layout -- 2026-03-13
  258. Heretic Defeats GPT-OSS with Arbitrary-Rank Ablation (ARA) Decensoring -- 2026-03-11
  259. Fat Fish — A Proper Upscale and Prune of Mistral Nemo -- 2026-03-11
  260. Optimizing Qwen3 Coder for RTX 5090 and PRO 6000 — Community Benchmarking Infrastructure -- 2026-03-11
  261. Yann LeCun's AMI Labs Raises $1.03 Billion to Build World Models -- 2026-03-11
  262. [Editorial] Bell Labs Solved Prompt Injection in 1976 -- 2026-02-27
  263. [Editorial] GitHub Copilot Exploited -- 2026-02-27
  264. [Editorial] Stop Telling Your AI Agent What Not to Do -- 2026-02-27
  265. [Editorial] AI MCP Integration Patterns -- 2026-02-27
  266. Turbo Connection: Reasoning as Information Flow from Higher to Lower Layers -- 2026-02-25
  267. O(1) Inference and Causal Monoid State Compression in Spartacus-1B -- 2026-02-25
  268. Sink-Aware Pruning for Diffusion Language Models -- 2026-02-25
  269. Liquid AI releases LFM2-24B-A2B -- 2026-02-24
  270. Qwen3's most underrated feature: Voice embeddings -- 2026-02-24
  271. After many contributions craft, Crane now officially supports Qwen3-TTS! -- 2026-02-24
  272. We tested the same INT8 model on 5 Snapdragon chipsets. Accuracy ranged from 93% to 71%. Same weights, same ONNX file. -- 2026-02-24
  273. [Editorial] https://www-cdn.anthropic.com/f21d93f21602ead5cdbecb8c8e1c765759d9e232.pdf -- 2026-02-12
  274. [Editorial] https://d3lm.medium.com/overly-agentic-why-anthropic-is-worried-about-opus-4-6-17eee0f8e5cd -- 2026-02-12
  275. [Editorial] https://www.linkedin.com/posts/avipil_i-got-my-first-bill-after-switching-to-claude-activity-7427320523870629889-vM5K -- 2026-02-12
  276. Pros/Cons and use case for bypassing permissions -- 2026-02-12
  277. [Editorial] https://www.kylerush.org/posts/opus-4-5-really-changed-things -- 2026-02-10
  278. [Editorial] https://www.linkedin.com/posts/ownyourai_i-taught-my-claude-code-to-swallow-a-32m-activity-7426902541868728321-2Z36 -- 2026-02-10
  279. [Editorial] https://docs.google.com/document/d/1I9r21TyQuAO1y2ecztBU0PSCpjHSL_vZJiA5v276Wro/mobilebasic -- 2026-02-10
  280. [Editorial] https://www.linkedin.com/posts/cole-medin-727752184_claude-codes-new-agent-teams-feature-is-share-7426633806792609792-pT3s -- 2026-02-10
  281. [Tool] claude-config-sync: Sync your Claude Code configuration across machines using GitHub Gists -- 2026-02-10
  282. [Editorial] https://www.linkedin.com/posts/ownyourai_i-just-woke-up-to-qwen3-coder-next-80b-activity-7424703876240695297-Nlqf -- 2026-02-04
  283. LiquidAI/LFM2.5-1.2B-Thinking -- 2026-02-04
  284. ByteDance-Seed/Stable-DiffCoder-8B-Instruct · Hugging Face -- 2026-02-03
  285. tencent/Youtu-VL-4B-Instruct -- 2026-02-03
  286. NousResearch/NousCoder-14B -- 2026-02-03
  287. transformers v5 final is out 🔥 -- 2026-01-30
  288. PaddlePaddle/PaddleOCR-VL-1.5 -- 2026-01-30
  289. zai-org/GLM-4.7 -- 2026-01-30
  290. MHA2MLA-VLM: Enabling DeepSeek's Economical Multi-Head Latent Attention across Vision-Language Models -- 2026-01-30
  291. Sharing my set of distilled small language models (3B) + training data in more than 50 low-resource languages -- 2026-01-30
  292. Introducing Kimi K2.5, Open-Source Visual Agentic Intelligence -- 2026-01-28
  293. ~60GB models on coding: GLM 4.7 Flash vs. GPT OSS 120B vs. Qwen3 Coder 30B -- your comparisons? -- 2026-01-28
  294. openbmb/AgentCPM-Report -- 2026-01-28
  295. Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice -- 2026-01-28
  296. Mixture-of-Models: Unifying Heterogeneous Agents via N-Way Self-Evaluating Deliberation -- 2026-01-28
  297. One Year Since the “DeepSeek Moment” -- 2026-01-28
  298. Qwen/Qwen-Image-2512 -- 2026-01-28
  299. [Editorial] https://www.linkedin.com/posts/ownyourai_deepseek-just-released-the-first-vision-ai-activity-7421818927657385987-V1yo -- 2026-01-27
  300. Unsloth announces support for finetuning embedding models -- 2026-01-27
  301. Qwen/Qwen3-TTS-12Hz-0.6B-Base -- 2026-01-27
  302. Qwen/Qwen3-VL-Reranker-2B -- 2026-01-27
  303. [Editorial] https://www.linkedin.com/posts/ivandj_early-claims-around-self-evolving-memory-activity-7421307316437676033-l0Jm -- 2026-01-26
  304. [Editorial] https://www.linkedin.com/posts/reuvencohen_introducing-ruvector-world-model-activity-7421556928910290944-cx4v -- 2026-01-26
  305. stepfun-ai/Step3-VL-10B -- 2026-01-26
  306. Qwen/Qwen3-VL-Embedding-2B -- 2026-01-21
  307. Phr00t/Qwen-Image-Edit-Rapid-AIO -- 2026-01-21
  308. Qwen/Qwen3-VL-Reranker-8B -- 2026-01-21
  309. Bartowski comes through again. GLM 4.7 flash GGUF -- 2026-01-21
  310. Step-Audio-R1.1 (Open Weight) by StepFun just set a new SOTA on the Artificial Analysis Speech Reasoning leaderboard -- 2026-01-20
  311. GLM-4.7-Flash -- 2026-01-20
  312. stepfun-ai/Step-Audio-R1.1 -- 2026-01-20
  313. tencent/HY-MT1.5-1.8B -- 2026-01-20
  314. Introducing GLM-Image -- 2026-01-14
  315. GPT-OSS -> MLA conversion breakthrough (20B), still looking for compute + collaborators -- 2026-01-14
  316. FrogBoss 32B and FrogMini 14B from Microsoft -- 2026-01-14
  317. Qwen/Qwen3-VL-Embedding-8B -- 2026-01-14
  318. MCP for Financial Ontology! -- 2026-01-09
  319. A community index for MCPs that don’t disappear after the thread ends -- 2026-01-09
  320. tencent/HY-WorldPlay -- 2026-01-09
  321. meituan-longcat/LongCat-Image -- 2026-01-09
  322. Introducing Falcon H1R 7B -- 2026-01-09
  323. [Editorial] https://www.linkedin.com/posts/ernst-van-gassen-9196a7b5_we-spend-a-lot-of-time-trying-to-make-prompts-ugcPost-7414966440023482368-dgh5 -- 2026-01-09
  324. [Editorial] https://leakhub.ai/ -- 2026-01-09
  325. [Editorial] https://www.linkedin.com/posts/hardmaru_survival-of-the-fittest-code-blog-https-activity-7415068590485458944-3tqP -- 2026-01-09
  326. A closer look at a BGP anomaly in Venezuela -- 2026-01-09
  327. Show HN: I visualized the entire history of Citi Bike in the browser -- 2026-01-09
  328. Modifying a QingPing Air Quality Monitor for Local MQTT Access -- 2026-01-09
  329. [Editorial] https://github.com/hiyouga/LlamaFactory -- 2026-01-07
  330. Tongyi-MAI/MAI-UI-8B · Hugging Face -- 2026-01-07
  331. MultiverseComputingCAI/HyperNova-60B · Hugging Face -- 2026-01-07
  332. [Editorial] https://www.alwaysfurther.ai/blog/train-4b-model-to-beat-claude-sonnet-gemini -- 2026-01-05
  333. [Experimental] Gemma 3 4B - Dark CoT: Pushing 4B Reasoning to 33%+ on GPQA Diamond -- 2026-01-05
  334. Youtu-LLM-2B-GGUF is here! -- 2026-01-05
  335. upstage/Solar-Open-100B -- 2026-01-05
  336. EditMGT — fast, localized image editing with Masked Generative Transformers -- 2025-12-30
  337. Francis-Rings/FlashPortrait -- 2025-12-30
  338. zai-org/GLM-TTS -- 2025-12-30
  339. Flowception: Temporally Expansive Flow Matching for Video Generation -- 2025-12-30
  340. I've been experimenting with SLM's a lot recently. My goal was to prove even SLMs can be accurate with the right architecture behind it. -- 2025-12-22
  341. MBZUAI releases K2-V2 - 70B fully open model. -- 2025-12-22
  342. microsoft/Fara-7B -- 2025-12-22
  343. Alibaba Tongyi Open Sources Two Audio Models: Fun-CosyVoice 3.0 (TTS) and Fun-ASR-Nano-2512 (ASR) -- 2025-12-19
  344. My professor lent me an A6000, so I tried to build a coding model. Here is Anni! (Qwen3-14B Fine-tune) -- 2025-12-19
  345. Two years ago, I was just a math major. Now I've built the 1.5B router model used by HuggingFace. Can I bring it to Cursor? -- 2025-12-19
  346. Independent evaluation of GPT5.2 on SWE-bench: 5.2 high is #3 behind Gemini, 5.2 medium behind Sonnet 4.5 -- 2025-12-16
  347. Did I overhype Claude Code? GPT + Comet are quietly beating it for me -- 2025-12-16
  348. 🚀 New: Olmo 3.1 Think 32B & Olmo 3.1 Instruct 32B -- 2025-12-16
  349. ByteDance/Dolphin-v2 -- 2025-12-16
  350. zai-org/AutoGLM-Phone-9B-Multilingual -- 2025-12-16
  351. [Editorial] https://www.linkedin.com/posts/eric-vyacheslav-156273169_a-7m-model-just-surpassed-deepseek-r1-gemini-activity-7405985266043297792-s1Jn -- 2025-12-15
  352. Nemotron 3 Nano \- A new Standard for Efficient, Open, and Intelligent Agentic Models -- 2025-12-15
  353. New in llama.cpp: Model Management -- 2025-12-12
  354. Successful prototype component prompt - For big projects -- 2025-12-12
  355. Building RNJ-1: What makes It different from Gemma 3? -- 2025-12-11
  356. EssentialAI/rnj-1 -- 2025-12-11
  357. From Azure Functions to FreeBSD -- 2025-12-10
  358. [Editorial] https://www.linkedin.com/posts/dakharlamov_too-slow-thats-what-they-called-engineers-activity-7403964370390646784-J27P -- 2025-12-09
  359. [Editorial] https://arxiv.org/abs/2511.20920 -- 2025-12-09
  360. [Editorial] https://www.linkedin.com/posts/anthony-alcaraz-b80763155_your-ai-agents-context-window-is-not-a-database-activity-7403394612893024256-9S87 -- 2025-12-09
  361. [Editorial] https://www.linkedin.com/posts/reuvencohen_introducing-ruvector-postgres-a-self-learning-activity-7403841311029837824-jogp -- 2025-12-09
  362. Emacs is my new window manager -- 2025-12-08
  363. shubh-io/DockMate -- 2025-12-08
  364. Comfy-Org/HunyuanVideo_1.5_repackaged -- 2025-12-08
  365. We were tired of guessing which local model to use for which query. built a speculative execution lib that figures it out (github) -- 2025-12-05
  366. Nimony (eventually Nim 3.0) Design Principles -- 2025-12-03
  367. nvidia/Orchestrator-8B · Hugging Face -- 2025-12-03
  368. allenai/Olmo-3-1125-32B -- 2025-12-03
  369. moonshotai/Kimi-Linear-48B-A3B-Instruct -- 2025-12-02
  370. orabazes/FLUX.2-dev-GGUF -- 2025-12-02
  371. Transformers v5: Simple model definitions powering the AI ecosystem -- 2025-12-02
  372. Pocketbase – open-source realtime back end in 1 file -- 2025-11-28
  373. [Editorial] https://www.rand.org/pubs/commentary/2025/11/electromagnetic-warfare-natos-blind-spot-could-decide.html -- 2025-11-28
  374. In depth analysis of Nvidia's Jet Nemotron models -- 2025-11-26
  375. Hidden causes of LLM latency, its not just the model size -- 2025-11-26
  376. Benchmark: Self-Hosted Qwen-30B (LoRA) vs. Llama-3.1-8B vs. GPT-4.1-nano. Comparison of parsing success rates and negative constraints. -- 2025-11-25
  377. [Editorial] https://arxiv.org/html/2510.04871v1 -- 2025-11-24
  378. [Editorial] https://www.linkedin.com/posts/ingason_agenticai-erp-aiagents-activity-7396551539538141185-PxNU -- 2025-11-24
  379. facebook/sam3 -- 2025-11-24
  380. rl-research/DR-Tulu-8B -- 2025-11-24
  381. Fast and Fluent Diffusion Language Models via Convolutional Decoding and Rejective Fine-tuning -- 2025-11-24
  382. marinero4972/Open-o3-Video -- 2025-11-17
  383. nanonets/Nanonets-OCR2-3B -- 2025-11-17
  384. nvidia/ChronoEdit-14B-Diffusers-Upscaler-Lora -- 2025-11-17
  385. ibm-granite/granite-4.0-h-350m -- 2025-11-14
  386. tencent/HunyuanWorld-Mirror -- 2025-11-14
  387. [Editorial] https://www.npmjs.com/package/neural-trader -- 2025-11-14
  388. Critical RCE patched in Imunify360 affects up to 50M+ websites -- 2025-11-14
  389. Kubernetes Ingress Nginx is retiring -- 2025-11-14
  390. About KeePassXC's Code Quality Control -- 2025-11-14
  391. Visualizing Quantization Types -- 2025-11-11
  392. Co-authored a book called "Build DeepSeek from Scratch" | Live Now -- 2025-11-11
  393. Writing your own BEAM -- 2025-11-11
  394. DIY Powerwall Blows Clouds, Competition Out of the Water -- 2025-11-11
  395. BAAI/Emu3.5-Image -- 2025-11-07
  396. Riemannian Optimization for LoRA on the Stiefel Manifold -- 2025-11-07
  397. "On-the-fly" code reviews with ollama. It kinda works.. -- 2025-11-07
  398. I Built an "AI Art Director" Agent to Orchestrate Image and Video Models. -- 2025-11-07
  399. What's your most unexpected Claude workflow discovery? -- 2025-11-07
  400. [Research] Cross-Stage Vulnerabilities in Large Language Model Architectures -- 2025-11-07
  401. runZeroInc/runZeroHound -- 2025-11-07
  402. openai/gpt-oss-safeguard-120b -- 2025-11-07
  403. Why does Image Recognition work in llama-server but not through Open WebUI? -- 2025-11-06
  404. Has anyone tested ollama on Whisplay HAT with Raspberry pi zero 2W? -- 2025-11-06
  405. allenai/olmOCR-2-7B-1025 -- 2025-11-06
  406. is there simple way like .bat to compress to q4-q8 like Unsloth, Qwen3-VL-30B-A3B-Thinking-abliterated model -- 2025-11-06
  407. MiniMax M2 Llama.cpp support merged -- 2025-11-05
  408. Which AI IDE should I use under $20/month? -- 2025-11-05
  409. lzA6/video-to-txt -- 2025-11-05
  410. xaviviro/python-toon -- 2025-11-05
  411. [Editorial] https://www.evokesecurity.com/blogs/prompt-injection-is-for-everyone -- 2025-11-04
  412. Found a remote file inclusion vulnerability in an AI-generated app before launch -- 2025-11-04
  413. Launch HN: Propolis (YC X25) – Browser agents that QA your web app autonomously -- 2025-11-04
  414. Attacking macOS XPC Helpers: Protocol Reverse Engineering and Interface Analysis -- 2025-11-04
  415. [Research] Unvalidated Trust: Cross-Stage Failure Modes in LLM/agent pipelines arXiv -- 2025-11-04
  416. [Editorial] Context Engineering Handbook -- 2025-11-03
  417. Qwen3-VL-32B Q8 speeds in llama.cpp vs vLLM FP8 on a RTX PRO 6000 -- 2025-11-03
  418. OSS alternative to Open WebUI - ChatGPT-like UI, API and CLI -- 2025-11-03
  419. Looking for a RAG UI manager to meet our needs to replace Zapier -- 2025-11-03
  420. LiquidAI/LFM2-1.2B-Extract -- 2025-11-03
  421. DeepSeek may have found a new way to improve AI’s ability to remember -- 2025-11-02
  422. Reliable Unlearning Harmful Information in LLMs with Metamorphosis Representation Projection -- 2025-11-02
  423. OpenAI: gpt-oss-safeguard: two open-weight reasoning models built for safety classification (Now on Hugging Face) -- 2025-10-31
  424. briaai/FIBO -- 2025-10-31
  425. snowyfizz/Vision-Detection-API -- 2025-10-29
  426. Show HN: Apache Fory Rust – 10-20x faster serialization than JSON/Protobuf -- 2025-10-29
  427. huggingface_hub v1.0: Five Years of Building the Foundation of Open Machine Learning -- 2025-10-29
  428. AlphaXiv,Compare the Deepseek-OCR and Mistral-OCR OCR models -- 2025-10-26
  429. Open-Bee/Bee-8B-RL -- 2025-10-26
  430. datalab-to/chandra -- 2025-10-26
  431. Unlock the power of images with AI Sheets -- 2025-10-26
  432. Picture in Picture / Webcam detect model on HuggingFace -- 2025-10-25
  433. Show HN: Story Keeper – AI agents with narrative continuity instead of memory -- 2025-10-25
  434. Show HN: Deta Surf – An open source and local-first AI notebook -- 2025-10-25
  435. Show HN: Tommy – Turn ESP32 devices into through-wall motion sensors -- 2025-10-25
  436. 20x Max Plan (€216) takes 2% of weekly Opus usage for a single Deep Research Query. That equals 50 per week if you use it ONLY for this and never continue or respond -- 2025-10-24
  437. I got tired of OpenAI dependency. Built a multi-LLM control center instead. -- 2025-10-19
  438. Turn ChatGPT into a real-time meeting assistant (via MCP + Apps SDK) -- 2025-10-19
  439. Claude Code taking a coffee break 🤔 -- 2025-10-19
  440. Show HN: Cmux – Coding Agent Multiplexer -- 2025-10-19
  441. [Editorial] Agentic Orchestration -- 2025-10-19
  442. Chicken Squisher 3000: Squish-Proof Security -- 2025-10-19
  443. Show HN: Largest open-source multimodal AI dataset -- 2025-10-18
  444. KORMo-Team/KORMo-10B-sft -- 2025-10-18
  445. ByteDance/FaceCLIP -- 2025-10-18
  446. Do you use multiple AI models for coding? Trying to validate a workflow problem -- 2025-10-17
  447. The “Compounding Engineering” mindset changed how I think about AI coding tools -- 2025-10-17
  448. Show HN: Rebuilt Bible search app to run 100% client-side with Transformers.js -- 2025-10-13
  449. swiss-ai/Apertus-8B-Instruct-2509 -- 2025-10-13
  450. A Childhood Dream, Created and Open Sourced -- 2025-10-13
  451. Some small tools for you - Ollama Managment UI, Passkey authentication proxy -- 2025-10-12
  452. Single prompt I run after git commit (before push) for AI diff/commit review -- 2025-10-12
  453. Show HN: Gitcasso – Syntax Highlighting and Draft Recovery for GitHub Comments -- 2025-10-12
  454. simonw/claude-skills -- 2025-10-12
  455. Custom models don't work after v0.6.33 update - Anyone else? -- 2025-10-12
  456. Pardus AI: Open source AI Assistant thanks for the help with Ollama -- 2025-10-09
  457. What to use for refactoring -- 2025-10-09
  458. Claude Code finally has a planning partner — I built an AI backlog manager for solo devs -- 2025-10-09
  459. Granite4 Small-h 32b-A9b (Q4_K_M) at FULL 1M context window is using only 73GB of VRAM - Life is good! -- 2025-10-09
  460. Run Open AI GPT-OSS on a mobile phone (Demo) -- 2025-10-09
  461. AI21 releases Jamba 3B, the tiny model outperforming Qwen 3 4B and IBM Granite 4 Micro! -- 2025-10-09
  462. inclusionAI/Ling-mini-2.0 -- 2025-10-09
  463. [Editorial] The Tiny Recursive Mode -- 2025-10-08
  464. deepseek-ai/DeepSeek-V3.1-Terminus -- 2025-10-08
  465. Behavioral Modification Systems in Large Language Models: A Methodological Analysis of Long Conversation Reminders -- 2025-10-08
  466. Ring Flash 2.0 104B A6B with Linear Attention released a few days ago -- 2025-10-07
  467. Bring Your Own Data (BYOD) -- 2025-09-30
  468. CohereLabs/command-a-reasoning-08-2025 -- 2025-09-30
  469. Built an MCP server for Claude Desktop to browse Reddit in real-time -- 2025-09-30
  470. MetalQwen3: Full GPU-Accelerated Qwen3 Inference on Apple Silicon with Metal Shaders – Built on qwen3.c - WORK IN PROGRESS -- 2025-09-28
  471. yangdongchao/UniAudio2 -- 2025-09-28
  472. jimsweb/aiMIDI -- 2025-09-28
  473. Handy – Free open-source speech-to-text app written in Rust -- 2025-09-28
  474. IndexTeam/IndexTTS-2 -- 2025-09-28
  475. beankeji-cloud/SLiteIO -- 2025-09-28
  476. GitHub - shantur/jarvis-mcp: Bring your AI to life—talk to assistants instantly in your browser. Zero hasle, No API keys, No Whisper -- 2025-09-27
  477. PC memory costs to climb as fabs chase filthy lucre in servers and HBM -- 2025-09-27
  478. AI and licensing (commercial use) -- 2025-09-26
  479. I trained an LLM from scratch AMA! -- 2025-09-26
  480. pengzhangzhi/Open-dLLM -- 2025-09-26
  481. SWE-Bench Pro -- 2025-09-23
  482. Investigating Training Data Detection in AI Coders -- 2025-09-23
  483. A first stab at packaging llama.cpp in a performance-optimized manner -- 2025-09-23
  484. Model: Qwen3 Next Pull Request llama.cpp -- 2025-09-23
  485. Unobtanium No More; Perhaps We Already Have All The Elements We Need -- 2025-09-21
  486. support for the upcoming Olmo3 model has been merged into llama.cpp -- 2025-09-21
  487. Running Nvidia CUDA Pytorch/vLLM projects and pipelines on AMD with no modifications -- 2025-09-21
  488. A Quick Look At The AMD Instinct MI355X With ROCm 7.0 -- 2025-09-21
  489. Uncensored AI model for from 4b Max 8b -- 2025-09-21
  490. facebook/MobileLLM-R1-950M -- 2025-09-20
  491. OpenGVLab/InternVL3_5-241B-A28B -- 2025-09-20
  492. KBlueLeaf/HDM-xut-340M-anime -- 2025-09-20
  493. GPT-OSS:20b & Qwen 4b are a match made in heaven for 24GB VRAM builds -- 2025-09-18
  494. Was working in RAG recently got to know how well Gemma3 4B performs -- 2025-09-18
  495. ROCm 6.4.3 -> 7.0-rc1 after updating got +13.5% at 2xR9700 -- 2025-09-18
  496. Is it possible for different brand GPUs to work together? -- 2025-09-18
  497. Beyond Prompt-Induced Lies: Investigating LLM Deception on Benign Prompts -- 2025-09-18
  498. LFM2-1.2B safety benchmark -- 2025-09-18
  499. Has anyone successfully gotten Ollama models (or any models) to execute SQL queries through natural language in Openwebui? -- 2025-09-18
  500. yangzhou24/OmniWorld -- 2025-09-18
  501. Analog Optical Computer for Inference and Combinatorial Optimization -- 2025-09-18
  502. Running Qwen-Next (Instruct and Thinking) MLX BF16 with MLX-LM on Macs -- 2025-09-17
  503. [Editorial] REFRAG: Rethinking RAG based Decoding -- 2025-09-17
  504. The Practicality Of Solar Powered Meshtastic -- 2025-09-17
  505. Kwai-Klear/Klear-46B-A2.5B-Instruct -- 2025-09-16
  506. WestZhang/VibeVoice-Large-pt -- 2025-09-16
  507. FlowSpec: Continuous Pipelined Speculative Decoding for Efficient Distributed LLM Inference -- 2025-09-16
  508. Qwen 3 Next Series – Qwen/Qwen3 Next 80B A3B Instruct Detected -- 2025-09-16
  509. Test-time Prompt Intervention -- 2025-09-16
  510. [Editorial] Tricks from OpenAI gpt-oss YOU can use with transformers -- 2025-09-15
  511. openbmb/MiniCPM4.1-8B -- 2025-09-15
  512. nunchaku-tech/nunchaku-qwen-image -- 2025-09-15
  513. ggml-org/gpt-oss-20b-GGUF -- 2025-09-15
  514. MBZUAI releases K2 Think. 32B reasoning model based on Qwen 2.5 32B backbone, focusing on high performance in math, coding and science. -- 2025-09-14
  515. unsloth/Qwen3-Coder-30B-A3B-Instruct-1M-GGUF -- 2025-09-14
  516. Effecient hot-swappable LoRA variant supported in llama.cpp -- 2025-09-12
  517. Qwen/Qwen3-Next-80B-A3B-Instruct -- 2025-09-12
  518. swiss-ai/Apertus-70B-Instruct-2509 -- 2025-09-12
  519. GRASPED: Graph Anomaly Detection using Autoencoder with Spectral Encoder and Decoder (Full Version) -- 2025-09-12
  520. stepfun-ai/step3 -- 2025-09-12
  521. Exploring State-Space-Model based Language Model in Music Generation -- 2025-09-12
  522. HuggingFaceModelDownloader v2.0 — fast resume, a slick TUI, and powerful filters for GGUF/variants -- 2025-09-12
  523. Roo Code Cloud is here with Task Sync & Roomote Control || Roo Code 3.28.0 Release Notes -- 2025-09-12
  524. Tricks from OpenAI gpt-oss YOU 🫵 can use with transformers -- 2025-09-12
  525. Made a one-click SearXNG fork with Redis, plus Dockerized Tika+OCR, and soon: local TTS/STT on Intel iGPU + AMD NPU -- 2025-09-12
  526. Mini-o3: Scaling Up Reasoning Patterns and Interaction Turns for Visual Search -- 2025-09-10
  527. Introducing FineVision: a huge open-source dataset for training SOTA Vision Language Models -- 2025-09-10
  528. wildminder/ComfyUI-VibeVoice -- 2025-09-10
  529. bytedance/USO -- 2025-09-10
  530. Wan-AI/Wan2.2-I2V-A14B -- 2025-09-10
  531. YanoljaNEXT-Rosetta: A Collection of Translation Models in Different Sizes -- 2025-09-07
  532. nasa-ibm-ai4science/Surya-1.0 -- 2025-09-06
  533. Vulkan back ends, what do you use? -- 2025-09-06
  534. A new OpenAI model? Could this be 5.1 or 5o? What do you think? -- 2025-09-06
  535. haasonsaas/dspy-0to1-guide -- 2025-09-06
  536. 16 reproducible failures → upgraded into a 300+ page Global Fix Map. one link inside, feedback wanted -- 2025-09-06
  537. VibeVoice RIP? What do you think? -- 2025-09-05
  538. lodestones/Chroma1-HD -- 2025-09-05
  539. Welcome EmbeddingGemma, Google's new efficient embedding model -- 2025-09-05
  540. NousResearch/Hermes-4-405B -- 2025-09-05
  541. After deepseekv3 I feel like other MoE architectures are old or outdated. Why did Qwen chose a simple MoE architecture with softmax routing and aux loss for their Qwen3 models when there’s been better architectures for a while? -- 2025-09-02
  542. The Hacker's Guide to Building an AI Supercluster -- 2025-09-02
  543. CAD, From Scratch: MakerCAD -- 2025-09-02
  544. PSO-Merging: Merging Models Based on Particle Swarm Optimization -- 2025-09-02
  545. Coral-Protocol/Anemoi -- 2025-09-01
  546. After researchers unmasked a prolific SMS scammer, a new operation has emerged -- 2025-09-01
  547. Silent No More: Open-Source Fix for Mic Mishaps -- 2025-09-01
  548. How to reliably detect cross-listed job ads across multiple sites? -- 2025-09-01
  549. GPT OSS Fine-tuning QAT -- 2025-08-31
  550. Semantic Structure in Large Language Model Embeddings -- 2025-08-31
  551. facebook/dinov3-vit7b16-pretrain-lvd1689m -- 2025-08-31
  552. MizzenAI/HPSv3 -- 2025-08-31
  553. Some thoughts on LLMs and software development -- 2025-08-30
  554. Show HN: Sideko – Hybrid deterministic/LLM generator for API SDKs and docs -- 2025-08-30
  555. [Editorial] AI interfaces for future -- 2025-08-29
  556. I’ve Debugged 100+ RAG/LLM Pipelines. These 16 Bugs Always Come Back. (70 days, 800 stars) -- 2025-08-29
  557. Updates to Consumer Terms and Privacy Policy -- 2025-08-29
  558. Hierarchical Reasoning Model (HRM) implementation for text generation -- 2025-08-27
  559. SQLite-Vector adds support for float16 and bfloat16 (CPU, NEON, AVX2 and SSE2) -- 2025-08-26
  560. zai-org/GLM-4.5 -- 2025-08-26
  561. rednote-hilab/dots.vlm1.inst -- 2025-08-26
  562. black-forest-labs/FLUX.1-Krea-dev -- 2025-08-26
  563. Datarus-R1-14B-Preview, an adaptive multi-step reasoning LLM for automated data analysis -- 2025-08-24
  564. Fully Open source, serverless, community-driven MCP alternative built in Python, TS and Go -- 2025-08-24
  565. unsloth/Kimi-K2-Instruct-GGUF -- 2025-08-24
  566. DeepSeek V3.1 Reasoner improves over DeepSeek R1 on the Extended NYT Connections benchmark -- 2025-08-24
  567. Qwen3-30B-A3B-Instruct 2507 vs Qwen3-Coder Flash -- 2025-08-23
  568. Your model zoo for Software dev / webdev -- 2025-08-23
  569. sriniously/go-boilerplate -- 2025-08-23
  570. Design Patterns in MCP: Literate Reasoning -- 2025-08-22
  571. FlyMyAI/flymyai-lora-trainer -- 2025-08-22
  572. Zedless: Zed fork focused on privacy and being local-first -- 2025-08-22
  573. Show HN: Rucat – Cat for Prompt Engineers -- 2025-08-22
  574. NEW VERSION: 0.6.23 Has Just Released! - Many fixes and new features, huge changelog -- 2025-08-22
  575. Docker Model Runner is really neat -- 2025-08-20
  576. Build a Powerful RAG Web Scraper with Ollama and LangChain -- 2025-08-20
  577. Ollama interface with memory -- 2025-08-20
  578. YuminosukeSato/pyproc -- 2025-08-20
  579. AGENTS.md – Open format for guiding coding agents -- 2025-08-20
  580. NVIDIA Nemotron Nano 2 and the Nemotron Pretraining Dataset v1 -- 2025-08-20
  581. deepseek-ai/DeepSeek-V3.1-Base · Hugging Face -- 2025-08-20
  582. mistralai/Devstral-Small-2507 -- 2025-08-20
  583. zai-org/GLM-4.5-Air -- 2025-08-20
  584. microsoft/Phi-4-mini-flash-reasoning -- 2025-08-20
  585. ModelTC/Qwen-Image-Lightning -- 2025-08-20
  586. Fast Type-Aware Linting in Oxlint -- 2025-08-20
  587. GPT-5, where does it shine for you? -- 2025-08-19
  588. vidore/colqwen-omni-v0.1 -- 2025-08-19
  589. Tutorial: Open WebUI and llama-swap works great together! Demo of setup, model swapping and activity monitoring. -- 2025-08-18
  590. Concurrency in open-weight/open-source models? -- 2025-08-18
  591. [Editorial] Claude Flow, Alpha 90 release -- 2025-08-17
  592. Optimizing Text gen webui (oobabooga) for MOE models (Qwen3-235b, GLM 4.5) -- 2025-08-17
  593. Chen-zexi/vllm-cli -- 2025-08-17
  594. zyfoxx/subhunter -- 2025-08-17
  595. HuggingFaceTB/SmolLM3-3B-Base -- 2025-08-16
  596. mistralai/Voxtral-Small-24B-2507 -- 2025-08-16
  597. TiTan - a tiny model for tags and titles -- 2025-08-16
  598. baidu/ERNIE-4.5-VL-424B-A47B-PT -- 2025-08-16
  599. MiniLM (BERT) embeddings in C from scratch -- 2025-08-16
  600. GENNAI CLI - A ReAct-based agent CLI -- 2025-08-16
  601. gptme v0.28.0 major release - agent CLI with local model support -- 2025-08-16
  602. WildFX: A DAW-Powered Pipeline for In-the-Wild Audio FX Graph Modeling -- 2025-08-16
  603. character-ai/pipelining-sft -- 2025-08-15
  604. Compass-Thinker-7B Technical Report -- 2025-08-15
  605. Can One Domain Help Others? A Data-Centric Study on Multi-Domain Reasoning via Reinforcement Learning -- 2025-08-15
  606. SaaS Is Dead -- 2025-08-15
  607. PYX: The next step in Python packaging -- 2025-08-15
  608. [Editorial] GLM-4.5, enterprise use -- 2025-08-13
  609. Best local model with function calling? -- 2025-08-13
  610. agentica-org/DeepSWE-Preview -- 2025-08-13
  611. janhq/Jan-v1-4B-GGUF -- 2025-08-13
  612. gpt-oss jailbreak workflow -- 2025-08-11
  613. GPT-5 removed logprob support from the API - technical breakdown and implications -- 2025-08-11
  614. A model for pure text continuation (not chirpy little Q&A assistant)? -- 2025-08-11
  615. Suggestion for upgrading hardware for MOE inference and fine-tuning. -- 2025-08-09
  616. Best models under 16GB?? -- 2025-08-09
  617. Explicit tail calls are now available on Rust Nightly (become keyword) -- 2025-08-09
  618. HuggingFaceTB/SmolLM3-3B -- 2025-08-09
  619. mistralai/Magistral-Small-2507 -- 2025-08-09
  620. N8N + OpenWebUI -- 2025-08-08
  621. Create space-saving clones on macOS with Python -- 2025-08-08
  622. Experience with GLM-4.5-Air + claude code? -- 2025-08-08
  623. Just when you thought Qwen was done... -- 2025-08-08
  624. Running GPT-OSS-120B at 500 tokens per second on Nvidia GPUs -- 2025-08-08
  625. Context Management by Trimming Conversation -- 2025-08-06
  626. Exploiting Primacy Effect To Improve Large Language Models -- 2025-08-06
  627. Help: Qwen3-Coder + LM Studio + Continue.dev (VSCode) + Mac 64GB M3 Max — 500 Internal Server Error, Even After Unsloth Fix -- 2025-08-04
  628. CWC now supports kimi.com (K2) and chat.z.ai (GLM-4.5) to enable coding with top tier models at no cost -- 2025-08-04
  629. 🧠 ICM+DPO: Used Qwen3's coherent understanding to improve Gemma3 at math - cross-model capability transfer with zero supervision -- 2025-08-03
  630. NoHumansRequired: Autonomous High-Quality Image Editing Triplet Mining -- 2025-08-03
  631. Build an AI Shopping Assistant with Gradio MCP Servers -- 2025-08-01
  632. [Editorial] Alternative to vector db rag -- 2025-07-30
  633. This year’s best open-source models and most cost-effective models -- 2025-07-29
  634. I’m looking for multimodal image input support and uncensored LLM -- 2025-07-29
  635. nvidia/audio-flamingo-3 -- 2025-07-29
  636. mistralai/Voxtral-Mini-3B-2507 -- 2025-07-29
  637. ziangcao0312/PhysX-3D -- 2025-07-29
  638. UI/UX benchmark update 7/22: Newest Qwen models added, Qwen3 takes the lead in terms of win rate (though still early) -- 2025-07-28
  639. unsloth/Qwen3-235B-A22B-Thinking-2507-GGUF -- 2025-07-28
  640. zai-org/GLM-4.5 -- 2025-07-28
  641. Tesslate/UIGEN-X-32B-0727 -- 2025-07-28
  642. had to fine-tune qwen since llama sucks at summarizing -- 2025-07-28
  643. orchestre-dev/ccproxy -- 2025-07-28
  644. [Editorial] neural networks don’t need to be giant to be powerful -- 2025-07-27
  645. Qwen/Qwen3-235B-A22B-Thinking-2507 -- 2025-07-27
  646. mistralai/Magistral-Small-2507 -- 2025-07-27
  647. cmdaltctr/claude-gemini-mcp-slim -- 2025-07-26
  648. RichardAtCT/claude-code-openai-wrapper -- 2025-07-26
  649. Freigeist - The new Vibe Coding Platform -- 2025-07-26
  650. From chaotic prompting to structured workflow: My Claude evolution -- 2025-07-24
  651. A Request for Comments (RFC) for MCP-alternative Universal Tool Calling Protocol (UTCP) was created -- 2025-07-22
  652. How to use the same context across LLMs and Agents -- 2025-07-22
  653. Has anyone actually ran VLAs locally and how good are they? -- 2025-07-22
  654. google/medsiglip-448 -- 2025-07-22
  655. Questions about AI for translation -- 2025-07-22
  656. microsoft/Phi-4-mini-flash-reasoning -- 2025-07-22
  657. LGAI-EXAONE/EXAONE-4.0-32B -- 2025-07-22
  658. Replacing thinking with tool usage enables reasoning in small language models -- 2025-07-22
  659. Struggling to Generate Polished UI with Claude Code -- 2025-07-20
  660. Skywork/Skywork-R1V3-38B -- 2025-07-20
  661. ByteDance-Seed/Seed-X-PPO-7B -- 2025-07-20
  662. Diffusion model support in llama.cpp. -- 2025-07-16
  663. GLM-4 MoE incoming -- 2025-07-16
  664. GeoArrow and GeoParquet, and the Future of Geospatial Data Analysis -- 2025-07-15
  665. Programming Affordances That Invite Mistakes -- 2025-07-14
  666. HuggingFaceTB/SmolLM3-3B-Base -- 2025-07-14
  667. OLMo 2 - a family of fully-open language models -- 2025-07-12
  668. google/videoprism -- 2025-07-12
  669. Why don’t we have a big torrent repo for open-source LLMs? -- 2025-07-12
  670. Local PDF Database searchable with ollama - best setup? -- 2025-07-12
  671. Tinyllama on old Mediatek G80 android device -- 2025-07-12
  672. I used Ollama to build a Cursor for PDFs -- 2025-07-12
  673. Advice on switching to LLM -- 2025-07-12
  674. Building the Hugging Face MCP Server -- 2025-07-11
  675. Support for the upcoming IBM Granite 4.0 has been merged into llama.cpp -- 2025-07-11
  676. support for Falcon-H1 model family has been merged into llama.cpp -- 2025-07-11
  677. fsndzomga/metadspy -- 2025-07-10
  678. osmosis-ai/Osmosis-Apply-1.7B -- 2025-07-10
  679. IntervitensInc/pangu-pro-moe-model -- 2025-07-10
  680. Continual Gradient Low-Rank Projection Fine-Tuning for LLMs -- 2025-07-10
  681. pola-rs/polars -- 2025-07-09
  682. Introducing an open source cross-platform graphical interface LLM client -- 2025-07-08
  683. llama-server vs llama python binding -- 2025-07-08
  684. Llama server completion not working correctly -- 2025-07-08
  685. introducing cocoindex - super simple etl to prepare data for ai, with dynamic index (ollama integrated) -- 2025-07-08
  686. Higher topk and num_ctx or map/reduce ? -- 2025-07-08
  687. Looking for an upgrade from Meta-Llama-3.1-8B-Instruct-Q4_K_L.gguf, especially for letter parsing. Last time I looked into this was a very long time ago (7 months!) What are the best models nowadays? -- 2025-07-08
  688. Best models by size? -- 2025-07-08
  689. Planning a 7–8B Model Benchmark on 8GB GPU — What Should I Test & Measure? -- 2025-07-08
  690. DLLM-Cache: Accelerating Diffusion Large Language Models with Adaptive Caching -- 2025-07-08
  691. Efficient MultiModal Data Pipeline -- 2025-07-08
  692. Are non-autoregressive models really faster than autoregressive ones after all the denoising steps? -- 2025-07-06
  693. Yuan-ManX/ComfyUI-OmniGen2 -- 2025-07-06
  694. Training and Finetuning Sparse Embedding Models with Sentence Transformers v5 -- 2025-07-06
  695. Using local models with Void -- 2025-07-05
  696. Intel GPU vLLM Docker Compose Bootstrap with Phi-lthy4 on A770 -- 2025-07-05
  697. Accelerated LLM Inference on AMD Instinct™ GPUs with vLLM 0.9.x and ROCm -- 2025-07-05
  698. Gemma 3n fully available in the open-source ecosystem! -- 2025-07-05
  699. 5060ti 16gb or 9060xt 16gb for small llm server -- 2025-07-05
  700. Qwen3 models in MLX format! -- 2025-07-05
  701. Best local coding model right now? -- 2025-07-05
  702. Found a Web3 LLM That Actually Gets DeFi Right -- 2025-07-03
  703. apple/DiffuCoder-7B-cpGRPO -- 2025-07-03
  704. Hoshinonyaruko/Gensokyo-MCP -- 2025-07-01
  705. ahmadallobani/BaldHead -- 2025-06-29
  706. modelcontextprotocol/registry -- 2025-06-27
  707. 0-concordance of knotted surfaces and Alexander ideals -- 2025-06-26
  708. unfinishedtr/progressbar -- 2025-06-24
  709. wearyfurnitur/confiq -- 2025-06-24
  710. Show HN: Controlling 3D models with voice and hand gestures -- 2025-06-24
  711. open-webui/mcpo -- 2025-06-23
  712. Built a fully local Whisper + pyannote stack to replace Otter. Full diarisation, transcripts & summaries on GPU. -- 2025-06-23
  713. MiniMax latest open-sourcing LLM, MiniMax-M1 — setting new standards in long-context reasoning,m -- 2025-06-23
  714. Run qwen 30b-a3b on Android local with Alibaba MNN Chat -- 2025-06-23
  715. A new PDF translation tool -- 2025-06-23
  716. What Really Happens When You Ask a Cursor a Question with GitHub MCP Integrated -- 2025-06-23
  717. [Q] How to Speed Up Mistral 7B Inference in LM Studio? 31s/Chunk on RTX 3070 -- 2025-06-23
  718. Cyber security guys are about to become very on demand in the coming few years -- 2025-06-23
  719. The first big AI disaster is yet to happen -- 2025-06-23
  720. Trading with Claude, and writing your own MCP server -- 2025-06-23
  721. Ecne AI Podcast Generator - Update -- 2025-06-23
  722. Help me decide on hardware for LLMs -- 2025-06-23
  723. chungmin99/pyroki -- 2025-06-23
  724. nvidia/GR00T-N1.5-3B -- 2025-06-23
  725. nvidia/Cosmos-Predict2-2B-Text2Image -- 2025-06-22
  726. trendmicro/vision-one-mcp-server -- 2025-06-20
  727. Minidoracat/mcp-feedback-enhanced -- 2025-06-20
  728. meta-llama/Llama-3.1-8B-Instruct -- 2025-06-19
  729. MiniMaxAI/MiniMax-M1-80k -- 2025-06-19
  730. ckanthony/openapi-mcp -- 2025-06-18
  731. Ta0ing/MCP-SecurityTools -- 2025-06-18
  732. fluxions/vui -- 2025-06-13
  733. carbon-language/carbon-lang -- 2025-06-12
  734. CJackHwang/AIstudioProxyAPI -- 2025-06-11
  735. News publishers call Google's AI Mode 'theft' -- 2025-06-11
  736. typelevel/cats -- 2025-06-11
  737. wesm/pydata-book -- 2025-06-11
  738. jedisct1/openapi-mcp -- 2025-06-09
  739. arcee-ai/Homunculus -- 2025-06-04
  740. GitHub MCP exploited: Accessing private repositories via MCP -- 2025-06-01
  741. dipampaul17/KVSplit -- 2025-06-01
  742. facebook/OMol25 -- 2025-05-30
  743. Is Microsoft’s new Foundry Local going to be the “easy button” for running newer transformers models locally? -- 2025-05-28
  744. Cobolt is now available on Linux! 🎉 -- 2025-05-28
  745. Round Up: Current Best Local Models under 40B for Code & Tool Calling, General Chatting, Vision, and Creative Story Writing. -- 2025-05-28
  746. Best local model for M2 16gb MacBook Air for Analyzing Transcripts -- 2025-05-28
  747. Prompt Debugging -- 2025-05-28
  748. Not so Smart Agent (Ollama, Spring AI, MCP) -- 2025-05-28
  749. [Q] How can one get better at fixing models,training etc.? -- 2025-05-28
  750. FOSS - MCP Server generator from OpenAPI specification files (swagger/etapi) -- 2025-05-28
  751. Digital Payment System GNU Taler Gets Green Light to Operate in Switzerland -- 2025-05-28