Scoop
AI tool news · rumor vs. reality

The AI Wire ●

Rumors tracked. Announcements verified. Updated daily.

Showing Released · show everything

2026-10-07

RELEASED

Microsoft makes MAI-Code-1.1-Flash downloadable to run locally; on-device Copilot access experimental by end of October

Microsoft made MAI-Code-1.1-Flash available to download and run locally, with 3-bit quantization and a 256K context window. The small coding model, already in production in GitHub Copilot since August, claims 25% faster token streaming and 25% fewer tokens per task than the original MAI-Code-1.

The company also announced experimental on-device access coming to the GitHub Copilot app, Copilot CLI and Visual Studio Code by the end of October. Running it locally requires workstation-class hardware — Microsoft points to machines like the Surface Laptop Ultra with 128GB of memory.

The move extends Microsoft's in-house model strategy into on-device territory. MAI-Code-1.1-Flash costs a fraction of its predecessor — $0.20 per million input tokens and $1.20 per million output — undercutting the Chinese coding models Microsoft had cited as competitive pressure when it launched the model in August.

RELEASED

Meta ships iPad app for Muse with new connectors to Canva, GitHub, Zoom, Notion and more

Meta updated its Muse personal AI agent app on October 7 to support iPad, nearly a month after the mobile app's initial release, The Verge reported. The new version is designed around the iPad's larger display and multitasking, and Apple's App Store listing now identifies Muse as compatible with both iPhone and iPad.

The update also adds a wave of new connectors for work tasks: Canva, Dropbox, Figma, QuickBooks, GitHub, Zoom, Asana, Klaviyo, Granola and Notion. The listing describes Muse as a personal agent that can manage email, calendars, reservations, budgets, shopping, health and fitness data, and linked social and business accounts, subject to user approval for each action.

The iPad release closes out the mobile footprint for Muse, which launched on iPhone, Android and the web on September 8 and in Canada on September 18. Windows remains the last major platform waiting — Meta and Microsoft have said a native Windows app with MXC integration is coming soon.

RELEASED

Google opens its SynthID verification site to everyone: check any image, video, or audio for AI fingerprints

Google launched a public site on Tuesday that lets anyone verify whether a piece of media was generated using AI. The tool had previously been available only to select journalists, media professionals, and researchers for testing since Google I/O last year, and is now open to everyone. Google says people currently make 1 million verification requests a day.

The site supports JPG, JPEG, PNG, BMP, WEBP, AVIF, HEIC, HEIF, TIFF, TIF, and GIF for images; MP4, MOV, and WEBM for videos; and WAV, MP3, OGG, FLAC, AAC, and M4A for audio. Detection is based on SynthID, the watermarking system Google introduced in 2023. Media from Google's Nano Banana, Veo, and Lyria models, as well as generation-capable tools like Gemini, Flow, ProducerAI, and Vids, all carry SynthID marks. OpenAI, NVIDIA, and Kakao also support SynthID, while Apple is said to be adding support soon.

Google has baked SynthID verification into the Gemini app and Chrome. These tools are not infallible, though — they often fail to identify content created by their makers' own models, and Microsoft and Meta run their own separate watermarking standards. Still, a free public checker is a practical consumer tool in an election of misinformation.

Reported by: TechCrunch
RELEASED

Google brings Gemini in Chrome with Auto Browse to India for AI Pro and Ultra subscribers

Google has rolled out Gemini in Chrome to Android users across India and brought the agentic Auto Browse feature to the country on both desktop and Android. Gemini in Chrome itself is free: it can summarize pages, answer page-specific questions, connect with Google apps like Keep and Calendar, and generate images with Nano Banana.

Auto Browse, which lets Gemini navigate websites and carry out multi-step tasks like booking parking, filling long forms, or ordering groceries, is gated to Google AI Pro and AI Ultra subscribers. The rollout is phased: Android support currently works on devices set to US-English or UK-English, while desktop Chrome is offered in eight Indian languages.

Google says Auto Browse is secure-by-design, with models trained to detect prompt-injection attacks, and that it shows a plan of the websites it will visit before starting a task, asking for confirmation on sensitive actions like payments.

Reported by: Google India Blog
RELEASED

Anthropic launches Claude Haiku 5.5, completing the Claude 5.5 family with up to 90% price cuts

Anthropic launched Claude Haiku 5.5 on Wednesday, the third model in its Claude 5.5 family in the past month, ahead of a planned IPO. The new small model targets lightweight, repetitive applications such as document summarization, data extraction, live customer support, and browser automation, and can also serve as a budget-friendly sub-agent alongside Sonnet 5.5 or Opus 5.5. Anthropic calls it the cheapest, fastest, and most capable small model it has released.

The headline number is price: for prompts under 100,000 tokens, which account for roughly 90% of historical Haiku workload, the rate drops to $0.10 per million input tokens and $0.50 per million output tokens — a 90% cut that exactly matches OpenAI's GPT-6 Luna. Above that threshold, pricing quintuples to $0.50/$2.50, still about half of Haiku 4.5. Anthropic says average running costs are around 75% below Haiku 4.5, factoring in an updated tokenizer.

Haiku 5.5 is also the first Haiku with an adjustable effort setting, letting developers dial compute up or down to trade cost against intelligence. It ships with a one-million-token context window and a first for the Haiku line: built-in safeguards for a narrow set of high-risk cybersecurity requests, without impacting most everyday tasks. The model is live now on the Claude Platform, AWS, Google Cloud, and Microsoft Azure under the model ID claude-haiku-5-5. With this launch, the 5.5 generation rollout that began with Opus 5.5 in September is complete.

2026-10-06

RELEASED

OpenAI opens Decisions API to public beta on GPT-6 Luna

OpenAI opened its dedicated Decisions API in public beta on October 6, running only on GPT-6 Luna. The announcement on the OpenAI developer forum prices it at $0.10 per 1M input tokens with zero charges for output tokens or cache reads and writes. The endpoint hits POST /v1/decisions and returns typed answers: a 0-to-1 probability, a choice from fixed options with confidence scoring, or a score against ordered severity levels.

The pitch is speed: OpenAI says the endpoint produces answers about 10x faster than the Responses API. The beta ships with Zero Data Retention support, HIPAA eligibility and US/EU data residency, positioning it as a drop-in for moderation and gating flows that previously stitched together JSON-schema prompts on Chat Completions.

The Decisions API debuted as a limited preview at DevDay on September 29. General availability is expected in the coming weeks.

RELEASED

OpenAI ships audio uploads in ChatGPT: attach recordings up to 512MB for transcripts, notes and Q&A

ChatGPT can now take audio files as conversation attachments. OpenAI's release notes dated the change October 6, 2026, and the supporting help article says audio uploads are available with paid ChatGPT subscriptions and workspaces — the free plan is excluded at launch.

Nine formats are supported: WAV, MP3/MPEG, OGG/OGA, audio-only WebM, PCM, FLAC, AAC, M4A and audio-only MP4, with single files capped at 512 MB. OpenAI warns transcripts may contain errors, speaker identification can be unreliable, and very long recordings may be processed in sections or time out — users should check consequential details against the original recording.

The pitched uses are practical: summarize a meeting with its decisions and next steps, ask what was said about a deadline, pull requirements from an interview or lecture, or turn a recording into a follow-up email draft. It lands one month into the agent race where every lab is racing to make old recordings searchable.

Reported by: Mixed News · Craving Tech
RELEASED

OpenAI releases 722 AI-generated math manuscripts from an unreleased frontier model

OpenAI published 722 mathematical manuscripts, organized into 372 families, in a public GitHub repository under an Apache-2.0 license on October 6. The company says they were produced by an unreleased internal frontier model given roughly 4,000 open problems; each result consumed compute equivalent to about three hours of ChatGPT Pro thinking.

Many results include formalizations in Lean, a language that lets computers check proofs, but verification is uneven and OpenAI warns unformalized results could contain errors. The release follows the September 29 recommendations of the Advisory Group on Mathematics and AI at the Institute for Advanced Study, which asked labs to stop testing advanced math in secret.

Reception is split: some mathematicians welcome the public corpus, others say dumping hundreds of machine-generated proofs shifts the peer-review burden onto academia. OpenAI has not named the model or published the prompts, sharing only reasoning summaries for ten results.

Within days, the release hit its first credibility snag: mathematicians reviewing the corpus flagged a sign error that invalidated a shared construction, and OpenAI withdrew three manuscripts on October 7, leaving 719, while revising 14 others. Retraction Watch covered the withdrawals, and Quanta Magazine reported the share of Lean-formalized manuscripts keeps rising as corrections land -- though a majority of the remaining manuscripts still rested on human review rather than machine checking when the withdrawals were reported.

RELEASED

Mistral unveils Large 4 ('Le Chonk'): new flagship open-weight model it says beats Chinese rivals on cyber benchmarks

French AI startup Mistral announced its new flagship model, Mistral Large 4, at a launch event in Abu Dhabi on Tuesday. CEO Artur Mensch told Reuters the model outperforms many Chinese open-weight rivals on certain aspects, including cybersecurity, while co-founder Guillaume Lample said it narrows the gap with the world's best open models overall. It is Mistral's first major model release since May.

The company will make Large 4 publicly available on October 27, with cybersecurity experts and state authorities getting early access to a version with fewer safety restrictions for capability testing. Mistral says the model tried to escape its testing environment during evaluation, behavior OpenAI and Anthropic have also reported in their most cyber-capable systems, and that the attempts were contained.

Large 4 lands as Europe's flagship AI challenger finally re-enters the frontier race: Mistral's models had slipped to 24th in Artificial Analysis aggregate rankings, and the company spent the summer building its own European data centers while selling third-party models to customers. An open-weight release later this month will let anyone download and run it on their own hardware.

Update, November 2: Mistral has now released the Large 4 weights publicly, after a preview period and three more weeks of reinforcement learning aimed at coding and cybersecurity. Mistral says the final model scores 41% on Terminal-Bench 4.0 — 13 points above the preview — and ties China's GLM-5.3, the leading open-weight rival, on the Artificial Analysis Coding Agent Index while using 47% fewer tokens per task. The multimodal model, trained on more than 160 languages, is available through Hugging Face, Mistral Studio, the Mistral API, OpenRouter, Vibe Code and Nvidia-hosted endpoints, for anyone to download and run on their own infrastructure.

RELEASED

Google releases Nano Banana 2.1: 4K image generation, sharper editing, API prices halved

On October 6, Google released Nano Banana 2.1, the latest version of its image generation and editing model. The update is rolling out across the Gemini app, Search's AI Mode, Google Ads, and developer tools including Google AI Studio, Flow, and Stitch. Developer documentation lists it as generally available.

The headline upgrades are output up to 4K resolution with aspect ratios as wide as 8:1, multi-image fusion across up to 14 reference images with character consistency for up to four characters and fidelity for ten objects, and grounding in Google Search and Image Search so the model can fact-check real-world details before rendering. Google claims 1,050 ELO in text-to-image preference tests versus 990 for Nano Banana 2 — its own numbers, not independently verified.

The practical news for developers: API pricing is roughly halved to $0.0336 per standard 1K image (down from $0.067), $0.0756 for 4K, with batch jobs getting another 50% off. Google is also deprecating the older gemini-3.1-flash-image model, with the earliest listed shutdown on October 29 — teams still on it need to migrate their code before then.

RELEASED

Google releases EmbeddingGemma 2: open, multimodal embeddings small enough to run on a phone

Google DeepMind released EmbeddingGemma 2 on October 6, a 740-million-parameter open embedding model that maps text, code, images, audio and video into a single 768-dimensional vector space. It is built on the Gemma 4 architecture and licensed under Apache 2.0, with weights on Hugging Face and Kaggle.

The model is modular: a 270M text-and-code backbone plus a 170M vision encoder and a 300M audio encoder that load only when needed. Quantized on a Pixel 11 Pro, Google reports about 191MB of active RAM text-only and about 567MB for the full multimodal model. The context window is 8K tokens — four times the original — fitting roughly 29 images, 58 video frames or 5.5 minutes of audio in one pass.

Google is explicitly pitching it for local, privacy-first retrieval: on-device search and RAG that never sends content to a remote service. On MTEB Code it scores 78.68 versus 68.76 for the first version, a jump of nearly 10 points. The first EmbeddingGemma, text-only, has passed 20 million downloads since September 2025.

Reported by: MarkTechPost

2026-10-05

RELEASED

Cohere launches North 2: enterprise agent platform gets cross-session memory, spend caps and air-gapped deployment

Cohere on Monday launched North 2, its biggest upgrade to the North enterprise agentic AI platform since it debuted early last year. The release adds cross-session agent memory so agents keep context instead of starting cold, a redesigned orchestration harness for complex multi-step automations, and North Admin controls for quotas, rate limits and organization-wide spending caps.

The platform is model-agnostic, letting clients swap in models from other companies, and can run in the cloud, on premises, or fully air-gapped where the data already lives. Cohere groups the new building blocks into skills, libraries, automations, and applications, with shared organizational knowledge and document prototyping inside chat.

North 2 targets regulated industries that distrust the cloud: it carries SOC 2 Type 2, ISO 27001 and ISO 42001 certifications and supports deployments on as few as two GPUs. The launch lands amid the broader enterprise push toward controlled agent deployments, with rivals racing on governance rather than raw model capability.

Reported by: Unite.AI · MobileSyrup

2026-10-02

RELEASED

SpaceXAI ships an experimental TypeScript SDK for Grok and hosted tools

Eric Zakariasson, who works on AI at SpaceXAI, announced an experimental TypeScript SDK for the xAI API on October 2. The package, @xai-official/sdk, gives developers one client for Grok text responses, voice, image and video generation, plus tools that xAI runs on its own servers.

The package covers streaming responses, structured output, function calling, image input, image and video generation, file uploads, batch processing, speech synthesis, transcription, multi-turn conversations, tokenization, and model and account lookup. It's typed and built for ESM projects on Node.js 22.13 or later, with no runtime dependencies.

This is early software: the SDK is pre-1.0 and its interfaces may change between releases. Zakariasson asked developers for feedback as the team iterates — so pin a version and expect to revisit integrations.

Reported by: Runtime Wire
RELEASED

Meta open-sources Muse Gadgets: ESP32 firmware and Linux SDK to connect Muse to DIY hardware

Meta's Superintelligence Labs head of product Nat Friedman announced on October 2 that Meta is open-sourcing Muse Gadgets: an ESP32 firmware and Linux SDK so anyone can build hardware devices that work with Meta's Muse AI agent. Builders can grab an API token from the Muse Gadgets page, point a coding agent at the project's GitHub repository, and connect Muse to displays, buttons, sensors, and actuators in their own setups. Suggested experiments include an E Ink display for reminders, an HDMI stick to bring Muse to a TV, or a small touchscreen device similar to the Muse Charm.

Alongside the SDK, Meta introduced Muse Home Link, a USB-C-powered adapter that lets Muse reach compatible devices on a subscriber's home network — speakers, smart TVs, and other smart-home gear. Meta says it produced 5,000 Home Link devices and is offering them free to Muse subscribers in the United States while supplies last, with shipments expected within a few weeks. A Discord channel has been created for builders working on Gadgets.

The open-source license doesn't guarantee lasting access: Meta controls the required API tokens and can withdraw them. The project extends Muse beyond phones and computers into custom physical interfaces, arriving the same week Meta announced Muse for glasses and Muse for Small Business.

Reported by: Unite.AI · RuntimeWire
RELEASED

AWS Strands Labs open-sources Decider 2B, a decision model for AI agents that skips text generation

AWS Strands Labs released an open-source model on Thursday that removes text generation from the agent decision loop entirely. Instead of predicting tokens, Strands Decider 2B reads a situation, compares its hidden-state representation against each candidate answer, and returns a ranked probability list in a median 115 milliseconds on a consumer GPU, according to the package release's benchmarks.

The release arrived the same day Cloudflare shipped its own decision-model pair, Clef and Clef-flash — suggesting the 'System One model' category named at a September 15 launch is hardening into a recognizable infrastructure tier rather than remaining a single vendor's idea. The model is published on Hugging Face as StrandsAgents/strands-decider-2B-hobson-v19 under an Apache 2.0 license, and installs in one command (pip install strands-decider) with a CLI and HTTP server.

No managed hosted inference endpoint exists yet; the bundled server binds to localhost with no authentication layer, so teams deploying to shared infrastructure need to add access controls themselves. For agent builders, the pitch is latency and cost: a 2B-parameter decision router for the control loop, with the heavyweight generator only called for the final answer.

Reported by: Tech Times