Daily AI news, sourcedUpdates / All
Updates Unconfirmed · 662 items · newest issue first · page 7 of 34
AreaAllModelsAgents & codingApps & productsChips & infraResearchPricing & limitsBusiness & fundingPolicy & safetyOther
- Dodaios, formerly Personal Jarvis, shows its open-source desktop agent app running two agents on local Ollama modelsUnconfirmedr/LocalLLaMAOct 13Apps & products
- SonicBoom, a Python-free C++/CUDA runtime, claims 52 tok/s decode on a laptop RTX 3050 versus 45 for llama.cppUnconfirmedr/LocalLLaMAOct 13Chips & infra
- Developer adds n+2 expert lookahead to SSD-streamed Qwen Flash Next on a 64GB Mac Mini M5Unconfirmedr/LocalLLaMAOct 13Chips & infra
- Developer has Claude Opus 5.5 add animated pixel-art support to its self-built SVG toolUnconfirmedr/singularityOct 13Apps & products
- Developer releases vramble, an open-source arbiter that gives local AI services one GPU lease at a timeUnconfirmedr/LocalLLaMAOct 13Chips & infra
- Blind quiz: 60% of respondents judged a Claude Opus 5.5 oboe concerto human-written, ahead of Handel and VivaldiUnconfirmedX @aug5thmusicOct 13Models
- User splits a 27B ternary model across two Android phones over USB, reporting about 2.8 tok/sUnconfirmedr/LocalLLaMAOct 13Models
- LocalLLaMA user starts a Qwen 3.8 Flash Next Q3_S benchmark after strong Exo results in OpenCodeUnconfirmedr/LocalLLaMAOct 13Models
- learnc tunes the constants in existing C code against data, cutting JPEG size by 9.1%Unconfirmedr/MachineLearningOct 13Agents & coding
- User reports running a 4-bit Qwen3.8-27B with 262k context and vision on a 16GB GPU at 80 tokens per secondUnconfirmedr/LocalLLaMAOct 13Models
- AskCyph Health Explorer pairs DeepMind's AlphaGenome with MedGemma 1.5 4B running in the browserUnconfirmedr/LocalLLaMAOct 13Apps & products
- Observer's developer shows local LLMs turning old laptops into home security monitorsUnconfirmedr/LocalLLaMAOct 13Apps & products
- Developer publishes a lecture and article on building a GPU compiler that lowers PyTorch graphs to CUDAUnconfirmedr/LocalLLaMAOct 13Chips & infra
- Codex CLI user asks OpenAI to make its persistent, self-updating background daemon opt-inUnconfirmedGitHub (openai/codex issue)Oct 13Agents & coding
- Precision Numerics releases Boole-1.7B, a 1-bit Qwen3-1.7B it says beats Bonsai-1.7BUnconfirmedr/LocalLLaMAOct 13Models
- EXL3-QLoRA brings native QLoRA training to ExLlamaV3 quants at 2-8 bpwUnconfirmedr/LocalLLaMAOct 12Agents & coding
- Open-source DeepSeek Harness plugin analyzes the quality of novel writingUnconfirmedr/LocalLLaMAOct 12Apps & products
- vLLM fork serves Qwen3.8-Flash-Next to 7 users at 262k context on six RTX 3090sUnconfirmedr/LocalLLaMAOct 12Models
- Bit-exact speculative decoding hits 106 t/s on Qwen3.8-27B with one RTX PRO 6000Unconfirmedr/LocalLLaMAOct 12Models
- Android receipt scanner runs a fine-tuned 0.8B vision model on-device for per-item grocery trackingUnconfirmedr/LocalLLaMAOct 12Apps & products