Home / New releases / Week of Nov 9–15, 2026

⚡ New AI tool releases: Nov 9–15, 2026

38 verified AI tool, model and product launches from the week of Nov 9–15, 2026, newest first. Each entry links to the official announcement or credible press; dates come from the vendor's post. Last updated 2026-11-13.

Friday, November 13

2026-11-13

Personal Agent Protocol (Poppy) v1.0 (Sierra)

Version 1.0 of Sierra's open standard for how personal AI agents interact with companies through business agents, websites and APIs. It is built largely on existing standards (MCP, OpenAPI and UCP), with OAuth so customers control what an agent can do. Shopify now publishes a Poppy file for every merchant automatically, Meta's Muse agent checks for Poppy first on every site it visits, and more than 100 companies are design partners. An implementation kit with a validator is available.

Source ↗
2026-11-13

Full-suite app API: web search, files, OCR, ASR and TTS (DeepSeek)

DeepSeek opened the capabilities behind its app to developers. V4.1-Flash (deepseek-flash) now supports server-side web search with citations, search during multi-step reasoning, file input (PDFs, Office files and images), automatic URL reading and safety checks in the Messages, Responses and Chat Completions APIs. Search, Fetch, OCR, speech-to-text and text-to-speech are also offered as standalone APIs; text-to-speech lists $0.45 per 10,000 characters (50% off for a limited time). Tool fees are charged on top of token prices. DeepSeek also open-sourced a demo chat client that can import DeepSeek chat history.

Source ↗
2026-11-13

llama.cpp v1.0.0 and ggml v1.0.0 (ggml-org)

The first stable major release of the widely used open-source library for running LLMs locally, about 3.5 years after the project began (the team is now part of Hugging Face). llama.cpp and ggml now follow semantic versioning: a major version bump means a breaking change to the public C API, minor releases add backward-compatible features, backends or model support, and patch releases are bug fixes. Builds are distributed for Linux, macOS, Windows, Android and iOS and through Docker, Homebrew, conda-forge, MacPorts and Winget.

Source ↗
2026-11-13

HeyGen Decision 1 (decision model API)

HeyGen's first decision model: one synchronous API call (POST /v3/models/decisions) reads text or JSON plus up to 4 images and one video (as sampled frames) and answers 1–400 named yes/no, choice and score questions, each with probabilities. It is aimed at classification, routing, triage and scoring content against your own criteria. heygen-decision-1 is available now (128K context per read); heygen-decision-1-turbo (256K, up to 100 images and 10 videos) is in preview. OpenRouter lists Decision 1 at $0.14 per 1M input tokens with no output charge.

HeyGen on SIloadout → · Source ↗

Thursday, November 12

2026-11-12

Gemini 4 Hydrogen (Google)

Google's fast Gemini 4 tier for everyday coding and agentic workflows. It is rolling out globally to every tier of Google AI Studio and the Gemini API, to Gemini for business, the Gemini app (free users get it through auto model selection) and Google Search, and to paid Google AI Pro and Ultra subscribers in Google Antigravity. It has four thinking levels (low, medium, high by default, and a new xhigh). Google says it beats Gemini 4 Argon on the Artificial Analysis Intelligence Index (51 vs. 50 at high, 54 at xhigh) and finishes tasks 2.5x faster than Gemini 3.8 Flash. Pricing is flat across the 1M-token context: $1.50 per 1M input and $7.50 per 1M output tokens. Google also introduced Orchestrator in Antigravity, which switches between Carbon and Hydrogen within a single session.

Source ↗
2026-11-12

Claude Meetings (Anthropic)

Claude can now join video calls as a visible participant. Teammates can ask it questions out loud or in chat (it answers in writing by default), and it can use connected tools with permission. It builds docs, whiteboards and other artifacts live while the team edits them, and it writes the notes and action items. Starting to roll out in beta on Team and Enterprise plans. Google Meet is supported first, with other video platforms coming soon. On Team plans, inviting Claude is on by default and an admin can turn it off.

Claude on SIloadout → · Source ↗
2026-11-12

Inkling 2 Small open weights (Thinking Machines Lab)

The full weights of Inkling 2 Small are now on Hugging Face under Apache 2.0, after external safety testing. It is a 594B-parameter mixture-of-experts model (17B active) with image input, variable effort and a 1M-token context, launched Nov 5 on Tinker with closed weights. It is integrated with Unsloth, SGLang, vLLM and TokenSpeed, can be post-trained with Miles, SkyRL and Snowflake Cortex Training, and is available by API from NVIDIA, Fireworks, Together and others.

Source ↗
2026-11-12

Atlas world model opens to everyone (World Labs)

World Labs opened its newest world model, Atlas, to everyone, with apps and API endpoints that turn ideas into keyframes, video clips or 3D Gaussian splats. Viewfinder reconstructs a scene from photos or generated images and renders camera paths and video. Chisel shapes a world through controllable scene building and grayboxing. Splats to Mesh converts splats into meshes for games and 3D tools, and Relight re-lights splats. A developer API and integrations, including Blender, let you build on it. No public pricing was listed at launch.

Source ↗
2026-11-12

UI SDK beta (Vercel)

An open-source (Apache 2.0) library of UI behaviors for coding agents, now in beta. Keyboard navigation, focus, selection, dismissal, popups, drag and drop, and the ARIA they need live in @ui-sdk/core, with adapters for React, Vue, Svelte, Solid and Web Components. Agents compose behaviors such as createPopup, createSelection and createTypeahead into components (a select, combobox or something new). The UI SDK agent skill installs with `npx skills add vercel-labs/ui`.

Source ↗
2026-11-12

Edge0-145B and Edge0-120B open weights (Edge0)

Open-weight on-device models for laptops, released with Edge0's unified inference framework and deployment tools. Edge0-145B is a sparse mixture-of-experts model with about 10B parameters active per step. Edge0 reports 29GB peak memory, 80+ tokens/s decode and a 128K–256K context on a compatible laptop, with day-one support for macOS, Windows and Linux (Apple M-series and NVIDIA RTX). The benchmarks are Edge0's own (e.g. 64.6 on DeepSWE 1.1). Weights are on Hugging Face (NVFP4 and q3k builds) under the Edge0 Community Model License 1.0, which is not a standard open-source license. An Edge0 App runs offline without an account.

Source ↗
2026-11-12

SOV-Transcribe-1 open weights (Salesforce)

Salesforce's first open-weight speech-augmented language model for non-streaming speech recognition, with 26B total and 4B active parameters. Weights are on Hugging Face under Apache 2.0. It targets best-in-class English transcription with multilingual support, entity understanding (phone numbers, emails), robustness to accents and noise, style steering and context biasing. Salesforce says it ranks #1 among open-weight models for non-streaming speech-to-text on Artificial Analysis (#7 overall) as of Nov 12, 2026.

Source ↗
2026-11-12

Dreamina Seedance 2.5 Flash (ByteDance)

A faster, cheaper version of the Seedance 2.5 video model. ByteDance says results are close to the flagship at nearly twice the generation speed. It outputs 480p or 720p only, with clips up to 30 seconds. BytePlus prices 720p text-to-video and image-to-video at USD 0.152 per second, which ByteDance puts at about 34% below Seedance 2.5. It is rolling out in Dreamina and partner apps such as Runway. API calls on BytePlus ModelArk opened as scheduled on Nov 15, 2026 at 14:00 (UTC+8).

Source ↗

Wednesday, November 11

2026-11-11

Aure voice translation model (DeepL)

DeepL's most lifelike Voice AI model, now powering DeepL Voice. It continuously adapts translated speech to the speaker's tone, emotion, emphasis and pace in real time, in their own voice, and it is never stored as a voice profile. Available across DeepL Voice (meetings in Microsoft Teams, Zoom and Google Meet, in-person conversations, and the DeepL API for Voice) in English, German, Spanish, French, Italian, Japanese, Portuguese and Korean.

Source ↗
2026-11-11

Gemini Decisions API and Gemini Decision 1 (Google)

A new Gemini API endpoint for fast closed-set decisions, powered by the new Gemini Decision 1 model (gemini-decision-1) in Google AI Studio. One multimodal input (text, images, audio, video or PDFs, up to 256K tokens) can be checked against up to 1,000 independent yes/no, multiple-choice (2–255 options) or rubric-score questions in a single pass, each returning probabilities and a 0–1 confidence score. Google quotes 150–250 ms end-to-end latency at $0.10 per 1M input tokens, with zero output tokens billed. Built for routing, policy checks, classification and grading.

Source ↗
2026-11-11

Grok Bot 1.0 (SpaceXSI)

Grok Bot leaves beta as shareable AI teammates: Team Bots can be shared with specific people or groups, work in Slack and Microsoft Teams, and join video calls, learning team context and preferences over time. Bots can also proactively offer to take on work and follow up on unfinished tasks, with better browser use, documents, slide decks and interactive Pages. Team Bots are on Team and Enterprise plans; Meetings are on Team, Enterprise and Ultra plans.

Source ↗
2026-11-11

New always-on Genie with Genie Ontology and Genie Search (Databricks)

Databricks turns Genie into an autonomous AI coworker for long-running enterprise jobs that runs on schedules or triggers, follows up on missing inputs and asks for approval where needed, reachable in Genie One, email, Slack and Microsoft Teams. It adds Genie Ontology (business context built by agents, governed in Unity Catalog), Genie Search across the lakehouse and apps like Google Drive, Jira and SharePoint, shareable artifacts, MCP connectors, plugins for ChatGPT, Claude, Copilot and Grok Bot, and Unity Gateway spend and access controls. Databricks enterprise customers only.

Source ↗
2026-11-11

AWS Emulator (Amazon Web Services)

A free local emulator for building and testing AWS applications on your computer or in CI without deploying or needing an AWS account: existing AWS SDK and CLI calls return in milliseconds, so developers and coding agents can find and fix errors offline. Services are generated from AWS's published API models; unsupported operations return 501 Not Implemented, and each instance gets its own port, keys and in-memory state. No service fee; source released under CC0 1.0.

Source ↗
2026-11-11

Tabletop (Google Labs)

A Google Labs experiment for real-time group collaboration: add links, files and ideas to a shared space, then prompt @tabletop (or pick a template) to generate interactive mini-apps such as trip itineraries, dinner polls, expense splitters and tournament brackets that everyone can use and edit at the same time. Rolling out via waitlist in the US to users 18 and older.

Source ↗
2026-11-11

ElevenAI early access and Interaction Engine preview (ElevenLabs)

At its NYC Summit, ElevenLabs opened free early access to ElevenAI, a workplace app with dictation and consent-based meeting notes (powered by Scribe v3), per-attendee action items and live in-call assistance. It also previewed Interaction Engine, which handles turn-taking and routes each part of a conversation to a cheaper or frontier LLM, and announced Memory and Routines for ElevenAgents, Scribe v3 Medical, Nemotron 3.5 Super VL in ElevenAgents, a Partner Program and the ElevenLabs Marketplace.

ElevenLabs on SIloadout → · Source ↗
2026-11-11

ERNIE 5.5 Flash Preview (Baidu)

Baidu's first post-trained model on the ERNIE 5.5 Flash Base: 280B total parameters with 14B active, a 1M-token context and over 200 tokens/s decoding, aimed at coding, 3D modeling, knowledge work and content creation. The API (ernie-5.5-flash-preview) is live on Baidu's Qianfan platform at a half-price launch promotion. Baidu calls it an intermediate checkpoint and says the weights will be open-sourced under Apache 2.0, with November 25 given as the date in its Chinese announcement.

Source ↗
2026-11-11

Qwen4-Max-Preview (Alibaba)

Preview of the next-generation Qwen flagship (model ID qwen4-max-preview): text, image and video input, reasoning on by default (low/medium/xhigh), tool calling, web search, structured outputs, a 1M-token context and up to 393,216 output tokens, compatible with the OpenAI and Anthropic APIs. Available only through Alibaba Cloud Model Studio's Token Plan (no pay-as-you-go); the preview runs Nov 11 to Nov 25, 2026 (UTC+8), after which this model is retired. No weights or benchmarks published.

Source ↗

Tuesday, November 10

2026-11-10

Gemini 4 Carbon (Google)

Google's strongest model, announced Oct 28 for cybersecurity defenders, is now generally available: built for long-horizon agentic work in software engineering, science and cybersecurity, with a 1M-token context and low/medium/high thinking levels. It is on the paid Gemini API tier in Google AI Studio and Gemini for business (gemini-4-carbon) at an introductory $2/$10 per 1M input/output tokens through Nov 30, then $4/$20, and rolling out to Google AI Ultra subscribers in the Gemini app, AI Mode in Search and Google Antigravity (with a new agent swarm preview). Gemini 4 Argon remains the default for most work.

Source ↗
2026-11-10

Claude Sheets, Whiteboards and Claude Data (Anthropic)

New Claude artifact types on all paid plans (web, desktop and mobile): Claude Sheets (beta) builds and co-edits spreadsheets with tabs, formulas and drop-downs, exportable to .xlsx/CSV; Claude Whiteboards (beta) turns team brainstorms into boards with timelines and can convert them to docs or decks; and Claude Dashboards becomes generally available as Claude Data, with certified queries and live sharing in beta. Interactive connectors now open partner apps such as Adobe, Atlassian, Box and Notion inside Claude. Anthropic says Claude Meetings (Claude joining calls as a participant) is coming soon.

Claude on SIloadout → · Source ↗
2026-11-10

Businesses in ChatGPT and Business Agents (OpenAI)

Beta that lets any eligible business claim its profile in ChatGPT through Business Hub, keep details and product feeds (pricing, availability) current, and see aggregated insights such as mentions, website clicks and discovery topics. Eligible US ecommerce and online-service businesses can also build Business Agents that answer customer questions in a clearly labeled conversation separate from ChatGPT's own answers, and can sponsor them as a destination for ChatGPT Ads.

ChatGPT on SIloadout → · Source ↗
2026-11-10

Adobe Creative Agent across Creative Cloud and Firefly Design (Adobe MAX)

At Adobe MAX, Adobe brought its Creative Agent (AI Assistants, beta) to Premiere, Illustrator, InDesign, After Effects and Frame.io (Photoshop via the Photoshop beta app) to run multi-step edits, and launched Firefly Design (beta), a single canvas for raster, vector, motion, 3D and print, plus Firefly Graph (beta) for repeatable multi-model workflows and Firefly Builder (private beta) for design-to-code prototypes. New tools include Relight and Generative Restyle in Photoshop and Color Mode and a Generative Media Tool in Premiere; local MCP integrations for third-party agents come in private beta in early 2027.

Source ↗
2026-11-10

MiniMax H3-Image open weights (MiniMax)

Open-weight image model built on MiniMax H3 for text-to-image, reference-based generation (up to 16 reference images) and editing, generating native 4K (4096x4096) images in under 10 seconds. Weights are on Hugging Face (MiniMaxAI/MiniMax-H3-Image) under a community license dated Nov 10 that excludes the EU, UK, South Korea and the US; it is also available through the MiniMax API, MiniMax Code and Hailuo AI, with 50% off 4K-and-above requests during launch week.

Source ↗
2026-11-10

Scribe v3 and Scribe v3 Live (ElevenLabs)

ElevenLabs' new speech-to-text models: Scribe v3 for recorded audio makes 27% fewer word errors than Scribe v2 across 102 languages with improved speaker diarization, and the low-latency Scribe v3 Live makes 43% fewer errors than Scribe v2 Realtime and adds speaker labels and context-aware turn detection. Both are in ElevenAPI (scribe_v3, scribe_v3_live) and are the default in the ElevenLabs app; v3 Live is also in ElevenAgents.

ElevenLabs on SIloadout → · Source ↗
2026-11-10

Nod (n8n Labs)

A research-preview personal agent from n8n Labs that runs on your own computer with its own local sandbox and browser, working in connected apps (Gmail, Calendar, Drive, Slack) and on websites with no integration: cancelling subscriptions, chasing invoices, watching prices or preparing a morning brief, on demand or on a schedule. It asks before it sends, pays, posts or deletes anything you didn't ask for. Free and open source (Apache 2.0) for macOS, with Windows 11 in preview; bring your own model provider. No n8n account needed.

n8n on SIloadout → · Source ↗
2026-11-10

Live 3 full-duplex voice model (Fish Audio)

Fish Audio's full-duplex voice model for voice agents keeps listening while it speaks, learning from real conversations when to acknowledge ("mm-hmm"), wait through pauses and corrections, or stop when a caller cuts in; it can decide to respond in as little as 186 ms. English only for now, on the Fish Agent Platform and API, with on-premises deployment for enterprises. Agent pricing is listed at $0.035/min (50% off $0.07) through Dec 31, 2026.

Source ↗

Monday, November 9

2026-11-09

Unreal Engine 5.9 MegaJam Preview with Unreal Assistant and Unreal Canvas (Epic Games)

Epic's preview build, which it calls the first Unreal Engine release built primarily with agentic workflows, adds experimental AI tools: Unreal Assistant (an in-editor panel for coding agents such as Claude Code, Codex, Cursor or OpenCode, connected to Unreal MCP), Unreal Canvas (a web app and editor tool that generates concept art, images, 3D meshes and video using your own model API keys), expanded Python APIs for agents and an experimental web (WebAssembly/WebGPU) target. Free download via the Epic Games Launcher and GitHub, timed for the Epic AI MegaJam on Nov 12–15; the full 5.9 release is planned for 2027.

Source ↗
2026-11-09

Claude Fable 5.2 (Anthropic)

Anthropic upgraded Claude Fable and Claude Mythos (the same underlying model; Fable adds stronger cybersecurity and biology safeguards so it can go to all users) with gains in coding, reasoning and vision, outputs up to 30% faster than 5.1, and a lower price of $8/$40 per 1M input/output tokens. An optional fast mode (up to 2.5x faster) costs $16/$80. Fable 5.2 is available on the Claude Platform (claude-fable-5-2), AWS, Google Cloud and Azure; Mythos 5.2 is limited to Anthropic's Cyber and Life Sciences Verification Programs and Claude Security for Enterprise.

Claude on SIloadout → · Source ↗
2026-11-09

Muse Ember (Meta)

Meta introduced Muse Ember, its smallest and most efficient Muse model, built for the work agents repeat all day: tool calls, coding subtasks, routing, extraction and long-document parsing. It takes text, image, audio and video input with a 1M-token context and up to 512K output tokens, and is the first model Meta trained using Muse Wonder Full. It is available as muse-ember-1.0 on the Meta Model API and Muse Code at $0.10/$0.50 per 1M input/output tokens ($0.004 cached input), with Responses API support, and Meta also brought it to Google Cloud.

Source ↗
2026-11-09

LongCat-2.5 open weights (Meituan)

One week after launching LongCat-2.5 (Nov 2), Meituan released its weights on Hugging Face under the MIT License, which allows commercial use, modification and redistribution. The natively multimodal mixture-of-experts model has 1.6T total parameters (about 48B active per token) and targets long-horizon agentic coding, visual coding and 3D tasks. A W4A4 quantized checkpoint is also published.

Source ↗
2026-11-09

Spacefast (Automattic)

Automattic launched Spacefast, a new company that hosts and publishes websites and apps built with the AI you already use (ChatGPT, Claude, Cursor, Claude Code, Codex and more) through official plugins, an API, CLI and MCP. Agents get expert build defaults and a ready database; work goes live in seconds on a private link teammates can comment on, and the agent turns comments into fixes. Free to start (personal plan with $1/month of usage, custom domains and a database); Go $5/mo; Pro $15/member/mo; Enterprise custom. App auth, transactional email and shared team knowledge are coming in the next few weeks.

Source ↗
2026-11-09

Emma and Nuplex Beta API (Nuance Labs)

Seattle startup Nuance Labs launched Emma, a face-to-face AI video persona built on Nuplex 1.0, its full-duplex audiovisual model that listens and speaks at the same time and uses visual cues such as nods and gaze for turn-taking (Nuance reports its reply timing overlaps a human's 89% of the time on 1,785 questions, versus 35% for the next system). Emma is free to try in the browser with no account; the Nuplex Beta API offers a few promptable personas that can hand longer tasks to your tools. English only, adults 18+, no face or voice cloning, watermarked API video; available in most of the US, the EU, the UK and Canada.

Source ↗
2026-11-09

ZONOS3-Mini open weights (Zyphra)

Zyphra released ZONOS3-Mini, a compact zero-shot voice-cloning text-to-speech model: a sparse mixture-of-experts with 100M active and 800M total parameters under Apache 2.0, with controls for identity, accent, mood and word-level delivery. It runs in real time on-device (under 150 ms to first audio on an M3 MacBook Air) with runtimes for browsers via WebGPU, SGLang/vLLM and GGML. Weights are on Hugging Face, and it is also served on Zyphra Inference.

Source ↗
2026-11-09

Seedream 5.5 (ByteDance)

ByteDance Seed launched Seedream 5.5, its next image creation model, with pixel-level stable multi-turn editing across up to 100 rounds, native RGBA output with transparent backgrounds, and more realistic images with clearer small and dense text. It is rolling out in Dreamina, CapCut, Doubao and via API on BytePlus and Volcano Engine Ark.

Source ↗

Frequently asked questions

What AI tools launched the week of Nov 9–15, 2026?

Personal Agent Protocol (Poppy) v1.0 (Sierra), Full-suite app API: web search, files, OCR, ASR and TTS (DeepSeek), llama.cpp v1.0.0 and ggml v1.0.0 (ggml-org), HeyGen Decision 1 (decision model API), Gemini 4 Hydrogen (Google), Claude Meetings (Anthropic), Inkling 2 Small open weights (Thinking Machines Lab), Atlas world model opens to everyone (World Labs), UI SDK beta (Vercel), Edge0-145B and Edge0-120B open weights (Edge0), SOV-Transcribe-1 open weights (Salesforce), Perplexity Decider v1.3 (open-weight decision models), Dreamina Seedance 2.5 Flash (ByteDance), Aure voice translation model (DeepL), Gemini Decisions API and Gemini Decision 1 (Google), Grok Bot 1.0 (SpaceXSI), New always-on Genie with Genie Ontology and Genie Search (Databricks), AWS Emulator (Amazon Web Services), Tabletop (Google Labs), ElevenAI early access and Interaction Engine preview (ElevenLabs), ERNIE 5.5 Flash Preview (Baidu), Qwen4-Max-Preview (Alibaba), Gemini 4 Carbon (Google), Claude Sheets, Whiteboards and Claude Data (Anthropic), Businesses in ChatGPT and Business Agents (OpenAI), Adobe Creative Agent across Creative Cloud and Firefly Design (Adobe MAX), MiniMax H3-Image open weights (MiniMax), Scribe v3 and Scribe v3 Live (ElevenLabs), Nod (n8n Labs), Live 3 full-duplex voice model (Fish Audio), Unreal Engine 5.9 MegaJam Preview with Unreal Assistant and Unreal Canvas (Epic Games), Claude Fable 5.2 (Anthropic), Muse Ember (Meta), LongCat-2.5 open weights (Meituan), Spacefast (Automattic), Emma and Nuplex Beta API (Nuance Labs), ZONOS3-Mini open weights (Zyphra), Seedream 5.5 (ByteDance).

How does SIloadout verify releases?

Each launch is checked against the vendor's own post, its official X account or credible press before it is added, and every source link is confirmed to load. Rumored, unreleased or ambiguously dated items are left out.

← Newer: Nov 16–22, 2026 · Older: Nov 2–8, 2026 → · All releases · Atom feed · JSON Feed

Browse by week: Nov 16–22, 2026 (25) · Nov 9–15, 2026 (38) · Nov 2–8, 2026 (20) · Oct 26–Nov 1, 2026 (30) · Oct 19–25, 2026 (45) · Oct 12–18, 2026 (33)

Browse by month: November 2026 (83) · October 2026 (108)

Open models only: New open-weight and open-source AI releases (33)

By company: Google (10) · OpenAI (7) · Microsoft (6) · Perplexity (5) · Tencent (5) · Anthropic (4) · ElevenLabs (4) · NVIDIA (4) · Baidu (3) · GitHub (3) · HeyGen (3) · Meta (3) · MiniMax (3) · Mistral AI (3) · Moonshot AI (Kimi) (3) · All companies