DoNews reports that OpenAI has reached an agreement with chip design software vendor Synopsys to jointly develop a chip design model called GPT-Synopsys. The report gives no details on the model's capabilities, release timing or the terms of the partnership.
At OpenAI DevDay 2026, OpenAI announced an upgrade to ChatGPT plugin extensions: developers can build in-app experiences in ChatGPT, give plugins a dedicated sidebar entry, build interactive panels for working alongside a conversation, or create viewers for supported file types, with plugin-building tools and a redesigned submission flow; OpenAI is also improving how plugins are sorted and recommended in the directory and in conversations, while users keep control over which plugins to use and approve access individually. Business, Enterprise, healthcare, and Education plans can connect supported ChatGPT plugins to a self-built site, letting team members sign in to the same app with their own permissions and their own connected data, and adding and managing automations has been simplified. OpenAI also said it is adding support for the proposed MCP event spec, so plugins can start automations when events occur in connected apps—for example, watching a project board for new tasks and drafting a plan from linked documents.
text-to-cad is an agent plugin that adds local CAD workflows: it generates 3D models as STEP, GLB, STL or 3MF from plain-language or image requests, runs design-for-manufacturing checks, produces dimensioned engineering drawings, and connects to 3D printing, sheet metal and CNC fabrication services. It installs into Claude Code, Claude Desktop, Codex, Cursor, Gemini and Grok Build — any agent supporting plugins or the skills framework — or as skills alone via npx skills add; it runs through uv and downloads the cadgen runtime on first start. Skills cover CAD, off-the-shelf STEP parts, engineering drawings, DXF, URDF/SRDF, SDF, SendCutSend pre-upload checks, DfAM printability checks, OrcaSlicer G-code and Bambu Lab printing.
Technology Innovation Institute (TII) has introduced Falcon-ASR, a 1.6B-parameter speech recognition model focused on Arabic, including the Emirati dialect. TII says it supports English, French, Spanish and Portuguese, offers word-level timestamps, and achieved a 20.92% average WER across six Arabic test sets in its evaluation.
Atomic Machines has emerged from six years in stealth with $250 million in funding and says its Matter Compiler can build working micro-machines directly from code. Its first product, the PrimeSwitch PS-150 relay, is aimed at protecting AI data centers transitioning to 800-volt direct-current power and is shipping to early-access customers for evaluation.
Nvidia explored a possible takeover of AI model marketplace OpenRouter but did not make a formal offer, according to a person familiar with the discussions. Stripe ultimately prevailed with an $8 billion bid, while Nvidia later pursued more than $140 billion in deals over two months, including a $105 billion credit guarantee.
Qualcomm has reportedly agreed to a multiyear cross-license with Huawei that may include patents supporting Huawei’s LogicFolding chip-stacking architecture, Bloomberg reported. The companies’ announcement did not identify the patents or mention LogicFolding; the deal also includes technology areas such as 5G, AI services, optics and networking, and remains subject to regulatory approval.
Microsoft CEO Satya Nadella and Nvidia CEO Jensen Huang are expected to present the Surface Laptop Ultra in San Francisco on Wednesday. The laptop is built around Nvidia’s RTX Spark chip and is designed to run AI agents locally, but its price, release date and final availability have not been announced.
Boston Dynamics has appointed former Amazon executive Rohit Prasad as CEO, effective October 7, succeeding Robert Playter, who stepped down in February. Prasad previously led Alexa and Amazon’s general artificial intelligence organization and oversaw development of the Amazon Nova foundation-model family.
The GitHub project ayghri/i-have-adhd offers a skill/plugin that claims to stop coding assistants from burying the answer: lead with the next action, number multi-step tasks, restate state every turn, end with one concrete next step, and suppress preamble and closers, with 10 rules listed in SKILL.md. Its README's before/after example contrasts a "Great question! Let me think about this..." explanation with direct instructions such as "Edit src/auth.ts:42" and "Run npm test -- auth.spec.ts." The project is MIT-licensed; the author says it is loosely based on The Adult ADHD Tool Kit by J. Russell Ramsay and Anthony L. Rostain, adapted for how an LLM should respond rather than how a person organizes their day, and users can fork it and edit SKILL.md.
San Francisco has temporarily banned new data centers within its borders, according to Bloomberg Technology. The city is home to some of the world's most valuable artificial-intelligence companies. The report did not specify the moratorium's duration or scope.
Anthropic is reportedly pursuing a custom AI accelerator designed by Broadcom and manufactured by Samsung's foundry on its SF2 2nm process, with an initial run said to be around 200,000 units. Rumored specs include roughly 4,000 TOPS INT8, about 1,300W TDP and 192GB of HBM3E (possibly 288GB), but Anthropic, Samsung and Broadcom have not publicly confirmed the arrangement or any of those figures. Anthropic has previously confirmed an internal chip team designing custom silicon for Claude, while saying its infrastructure will keep using AWS Trainium, Google TPU and NVIDIA GPUs, and has announced expanded Google and Broadcom partnerships for multi-gigawatt TPU capacity starting in 2027.
OpenMontage has appeared on GitHub as a self-described first open-source, agentic video production system, bundling 12 production pipelines, 100+ tools and 700+ agent skill and production-knowledge files. It plugs into AI coding assistants such as Claude Code, Cursor, Copilot, Windsurf and Codex, letting an agent handle research, scripting, asset generation, editing and rendering using tools including Remotion, Blender, FFmpeg and Piper TTS. A local Backlot board turns the asset-generation contact sheet into an approval gate before the final render.
Cognee is an open-source AI memory platform that gives agents persistent long-term memory across sessions, turning documents, code and conversations into a self-hosted, searchable knowledge graph that can run locally with small models and no cloud LLM key. Per its release notes, v1.6.1 (September 24) bundles Google Drive and Gmail OAuth connectors into the SDK, streams visualization of large graphs in chunks, and moves the GLiNER installer off the event loop; v1.6.0 supports keyless memory building and retrieval. The project reports BEAM conversational-memory scores of 0.79 at 100K tokens and 0.67 at 10M tokens, while noting the two settings use different conversations, ingestion models and retrieval procedures.
An NVIDIA technical blog describes using the node-based NVIDIA DSX Air digital twin to validate supported configurations and software integrations for AI factories before hardware arrives. Agents can query the twin, run configuration, security and compliance checks, compare results against organizational policy and return evidence-backed recommendations inside a governed automation loop with human-in-the-loop gates. NVIDIA Brev supplies on-demand GPU compute attached to the twin, and the NVIDIA AI Blueprint for Video Search and Summarization is used as an example of agent-orchestrated video analysis, retrieval-augmented knowledge and report generation routed to systems such as Jira.
Google signed a 20-year power purchase agreement with Constellation Energy to support 890MW of new nuclear generating capacity by uprating six existing plants. Constellation expects to invest more than $4.3 billion in the program, with the first uprated plant due online in 2028 and the full 890MW delivered by the end of 2032. Google said the deal brings its total enabled new nuclear capacity in the US to over 1.5GW; the report notes the power is not exclusively destined for Google's data centers but will support their expansion.
Anthropic has open-sourced a set of role-based plugins on GitHub, with an initial 11 covering sales, customer support, product management, marketing, legal, finance, data, enterprise search and bio-research. Built for Claude Cowork and also compatible with Claude Code, each plugin bundles skills, connectors, slash commands and sub-agents as plain markdown and JSON files with no code, infrastructure or build steps. Teams can swap connectors, add company terminology and processes, or write new plugins following the same structure.
According to publishing-industry sources, Nippan Group Holdings — parent of major Japanese book wholesaler Nippan — confirmed in a letter to the Japan Book Publishers Association that it sold a large quantity of books to US AI company Anthropic. The transaction was reportedly confirmed recently. Neither the letter nor the report disclosed the number of books, the value or the intended use.
OpenAI announced the Decisions API on its API platform, which evaluates text, images or both and is said to run about ten times faster than the Responses API. It is in public beta and currently supports only gpt-6-luna, priced at $0.10 per million input tokens with output tokens free, returning yes/no probabilities, picks from predefined categories, or scale-based ratings. OpenAI also cut its paid API tiers from five to three—Build, Launch and Grow—with monthly usage limits of $500, $5,000 and $200,000.
Tradewinds, a procurement marketplace run by the Pentagon's Chief Digital and Artificial Intelligence Office (CDAO), lets companies submit product videos of no more than five minutes describing how their AI offerings address periodically updated "strategic focus areas"; a panel reviews submissions at least monthly, and accepted products gain "post-competitive" status that makes it easier to buy them via vehicles such as Other Transaction agreements. A DOD official says some awards have been made in under a week, and OpenAI, Anthropic and Google are all listed as participants. WIRED reports that 2026 topics seek AI that can "enhance force lethality" and "unleash" AI-enabled agents for joint targeting, including so-called kill chain execution, while earlier pitches on a "culture of responsible AI" and guardrails are no longer sought.