Daily AI news, sourcedUpdates / All
Updates Official · 968 items · newest issue first · page 18 of 49
AreaAllModelsAgents & codingApps & productsChips & infraResearchPricing & limitsBusiness & fundingPolicy & safetyOther
- Eight US states sue OpenAI, alleging ChatGPT is unsafe for children and was marketed misleadinglyOfficialIowa Attorney GeneralOct 15Policy & safety
- Microsoft Foundry updates GPT-chat-latest with GPT-6-level intelligence, GenUI interactive answers and interleaved-thinking outputOfficialMicrosoftOct 15Models
- Suno explains sidechain compression in Suno Studio, whose browser-based compressor can use any track in a project as the sidechain input
- Microsoft Research releases UserRepGen-4B, an MIT-licensed model that turns conversations into task and behavior descriptions for user simulatorsOfficialHugging Face @microsoftOct 15Models
- GitHub updates Copilot's HydraFusion multi-model orchestration and adds HydraFusion Max for harder tasks
- xAI releases Grok Voice 3.0, which tops Artificial Analysis's speech-to-speech index, in its Voice Agent API at Grok Voice 2.0's $0.08-a-minute priceOfficialxAIOct 15Models
- Budget-matched 15M decision models show a linear-recurrent backbone running 2.37x faster than attention at 8K context, with accuracy near chanceOfficialGitHub @stube123890-hueOct 15Models
- TheoPhysEval: GPT-6 Astra in Codex solves 49 of 50 unpublished research-level theoretical physics problems at under $3 eachOfficialarXivOct 15Models
- Telperion releases Qwen3.8-27B-MLA and Qwen3.8-27B-GLA-g2, training-free attention retrofits that halve the model's KV cacheOfficialHugging Face @TelperionAIOct 15Models
- Kokon, a free open-source desktop app, runs Qwen-Image 2.1 photo editing locally on GPUs with 12 GB or more and no setupOfficialGitHub @GabrieleLozuponeOct 15Apps & products
- Zellige AI releases Tessera-27B-B-CJK, a Qwen3.8-27B coding-agent fine-tune that thinks in compressed CJK shorthand to use about half the thinking tokensOfficialHugging Face @ZelligeAIOct 15Models
- Mr.Chatterbox-2.7B-Chat, a model pretrained from scratch only on 19th-century British text, is fine-tuned to converse as a Victorian gentlemanOfficialHugging Face @tventurellaOct 15Models
- AWS shows a self-improving document processing system of eight agents on Bedrock AgentCore that learns each new document format onceOfficialAWSOct 15Agents & coding
- OpenAI says SentinelOne uses OpenAI Daybreak to reverse-engineer malware in about eight hours instead of four to six weeks
- OpenAI says its models account for most tokens used in Amp, which uses GPT-6 Astra, Ultrafast and GPT-Live-1 for agent work beyond codeOfficialOpenAI NewsOct 15
- Moonshot AI's Kimi Code Desktop 1.1.0 adds a Linux version and Remote Control for steering the local agent from a phone or another computer
- llama.cpp merges a chunked CUDA/HIP kernel for GDN layers that speeds up Qwen3.5-35B-A3B prompt processing by up to 21%OfficialGitHub (ggml-org/llama.cpp)Oct 15Agents & coding
- inclusionAI open-sources Ling-3.1-flash under MIT, a 560B-total, 25B-active hybrid reasoning model with a 1M-token context window
- Ai2 releases HiRO-ACEv2, which generates global 3 km climate data at about 1/300th the cost of the X-SHiELD storm-resolving modelOfficialAi2 (Hugging Face Blog)Oct 15Models
- NVIDIA sets October 15 release for Nemotron 3.5 Bolt 74B-A9B, an open MoE model packed into 24GB at 2 bits for local agents on RTX Spark