Products

ByteDance and Tencent are speeding up personal AI agents to compete after Meta's Muse took off, Caixin reported on Oct. 12. Citing a company source, Caixin says ByteDance accelerated before the National Day holiday, building on an April project code-named Spell for its Doubao assistant. The report is paywalled; Tencent's plans are not described in the portion we could read.

At Gemini at Work 2026 on Oct. 8, Google Cloud announced the Gemini agent, which it calls a universal agent for work: it takes objectives, plans tasks, uses tools and connects to business systems including Microsoft 365, Slack and Jira, and can run on Anthropic's Claude as well as Gemini. A Google Cloud spokesperson told VentureBeat it is in private preview with selected customers, with no date for wider availability. Businesses come first and consumers later. Capability claims are Google's; no independent benchmarks were published.

OpenAI began rolling GPT-6 into ChatGPT on Oct. 7: GPT-6 Sol for Plus, Pro, Business and Enterprise from Oct. 7, GPT-6 Luna for Free and Go from Oct. 8. The models reached the API on Sept. 22, so this is a distribution event, not a model launch. The new piece is 'Intelligent UI', which builds charts, forms, buttons and small in-chat tools into answers. OpenAI's speed and quality figures come from its own internal evals and have not been independently tested in ChatGPT.

Elon Musk posted that xAI's Grok Bot will use the 'best back end model for any given task,' naming Anthropic's Claude Opus 5.5 (as relayed), Midjourney and Suno — 'Whatever is most likely to give you the best outcome,' per City AM. The post describes a direction, not a documented launch or partnership; neither xAI nor Anthropic has published commercial terms, and the post did not specify routing, pricing, user controls or data handling.

Meta and Bret Taylor's Sierra, with Walmart, Stripe, Shopify, Genesys, Instinct and Rocket, announced Personal Agent Protocol — an OAuth-based open standard for how consumer agents sign in and act on business sites. There is no published spec yet; Sierra plans v0.1 later this month. OpenAI and Anthropic are not part of it for now; Taylor expects them later. TechCrunch also lists NiCE and Decagon among partners.

Google launched Playground, an experimental platform where US users 18 and older describe a game in text, play it immediately and change physics, rules or characters by asking — with tiered creation access by Google One membership. Unity Spark, which adds pro mechanics and the Unity runtime, is in testing with a closed beta coming soon. Roblox shares fell as much as 8% in premarket trading after the announcement, per The Next Web citing Investing.com; closing move not checked. Models named via a Google spokesperson to The Verge, as relayed by TNW.

Google said Gemini in Chrome is now available to all Android users in India, and its agentic auto browse feature is open on desktop and Android for AI Pro and AI Ultra subscribers there. Availability starts in US/UK English; auto browse is limited to English-set devices. Google calls the features secure-by-design. This is a geographic rollout of an existing capability, not a new agent framework.

Anthropic launched Claude for Google Workspace in public beta on all paid Claude plans: a Marketplace add-on that opens a Claude sidebar in Docs, Sheets and Slides, plus connectors to create and edit Google files from chat. Default ‘Ask before edits’; optional ‘Accept all’. Enterprise CMEK, Compliance API and OpenTelemetry audit export apply. Gemini is already native in Workspace. xAI shipped a Grok Workspace add-on in July — this is not a first.

Meta’s Muse Gadgets page says ESP32 and Linux device SDKs and firmware are open source under Apache 2.0 so builders can connect Muse to their own boards, including a Raspberry Pi. Muse Home Link — a Meta-made Wi-Fi bridge to local HTTP devices — is free with an active U.S. Muse subscription, limit one, first come first served, and ships in October. The Verge reports Nat Friedman said Meta manufactured 5,000 Home Links; that unit count is not on the product page.

OpenAI began rolling out Health in ChatGPT to logged-in U.S. users 18+ on Free through Pro—medical records, Apple Health, One Medical, Function Health—one day after a California suit sought to block the product over alleged dangerous medical advice. Free-tier medical-record context meets live liability testing.

OpenAI’s release notes restored ChatGPT on WhatsApp in the EEA—text, images, voice notes, optional account linking for higher limits—after Business Solution policy had pushed third-party AI bots off. Distribution win under GDPR/DMA pressure; not a global re-default of all bots.

TechCrunch: OpenAI is hiring a San Francisco Product Manager for families, caregivers, and older adults as ChatGPT’s 35+ share rises and nearly one in four U.S. parent smartphone users are on the app—amid teen-safety lawsuits and parental-control rollouts.

Three days after Muse Image launched, Meta pulled Instagram’s @-mention public-account reference feature after privacy and likeness backlash—default opt-in for public profiles. Muse Image remains on Meta AI and WhatsApp; social-graph distribution failed the consent test.

OpenAI launched GPT-Live-1 and Live-1 mini for ChatGPT Voice—full-duplex listen-and-speak on consumer web/mobile; Business/Enterprise/Edu excluded at launch. Mass-user agent UX bid while governance surfaces wait.

xAI shipped 21 new multilingual flagship Grok voices across Voice Agent API, TTS API, and Voice Agent Builder—plus naturalness upgrades to the original five. Platform plumbing for casting, not one default robot; same week as GPT-Live.

xAI launched the Grok Voice Agent Builder on its console—configure and ship voice agents on the realtime Voice Agent API and TTS stack. Builder first (July 1), 21-voice catalog days later, GPT-Live same week: productizing voice agents, not only chat.

OpenAI shipped a first-class Scheduled page in ChatGPT for Plus/Pro/Business/Enterprise—view, pause, edit recurring jobs and monitors. Makes lightweight automation legible and revocable as memory, apps, and connectors expand.

Microsoft made Copilot Cowork generally available worldwide for Microsoft 365 Copilot customers after Fortune 500-heavy preview. GA is when agent features stop being demos and start being credits, governance, and support SLAs.

Anthropic paused the June 15 plan to move Agent SDK, headless claude -p, and third-party apps off Pro/Max subscription pools onto separate API-priced credits. Temporary relief for builders; industry still negotiating who pays for unattended agents.

Enterprises can draw Oracle Universal Credits for OpenAI models and Codex on OCI. Frontier model distribution becomes a multi-cloud commitment-pool product—pricing power and CIO leverage shift accordingly.

OpenAI broadened Lockdown Mode across personal ChatGPT and self-serve Business—opt-in that disables browsing and outbound agent tools against prompt-injection exfiltration. Acknowledges residual risk in cached content and uploads; security as capability vs containment spectrum.

OpenAI began rolling Dreaming V3 to U.S. Plus/Pro—automatic background synthesis of ChatGPT memory, ~5× more compute-efficient than prior designs. Path to Free-tier personalization; sharper questions on profile opacity and consent without explicit saves.

OpenAI’s Codex gains Windows Computer Use (see/click/type in apps), remote steering from mobile ChatGPT or Mac, and Profiles for usage visibility — while EEA, UK, and Switzerland are carved out at launch. Agentic coding moves from repo edits toward full desktop control where policy allows.

xAI’s Grok Build early beta brings a terminal-native coding agent to SuperGrok and X Premium+ — plan mode, parallel subagents/worktrees, AGENTS.md/MCP fit, headless automation. Coding agents are now table stakes for pro subscriptions, not only enterprise APIs; no independent Elo is published with the launch.