As of October 13, 2026.
The big AI companies spent the week pushing their assistants further toward doing work, with answers that include working tools and agents that return finished documents.
At a glance
| Item | What it is | Who it's for | Our take |
|---|---|---|---|
| GPT-6 and Intelligent UI in ChatGPT | New models in ChatGPT, with forms, charts and calculators inside answers | Anyone already using ChatGPT | Try it, low cost to test on a small task |
| Gemini agent (Google Cloud) | A work agent that plans tasks and returns finished work in documents and email | Businesses on Google's business tools | Watch it, depends on how your accounts are set up |
| Muse Wonder and Muse Wonder Full (Meta) | Meta's most capable models for agent tasks | Developers and Muse Code subscribers | Skip it for now, built for builders |
| Mistral Large 4 preview | A large multimodal model in public preview, open weights promised | Developers and technical teams | Watch it, wait for the open weights |
OpenAI brings GPT-6 and Intelligent UI to all ChatGPT users
On October 7, OpenAI began rolling out GPT-6 in the ChatGPT Chat tab, starting with Plus, Pro, Business and Enterprise. It expanded to Free and Go the next day. Paid tiers use GPT-6 Sol and Free and Go use GPT-6 Luna. OpenAI says the models behind ChatGPT Work and Codex are not changing with this release, and Enterprise availability depends on admin settings.
A new feature called Intelligent UI lets answers include visuals, buttons, forms, charts and small interactive tools, such as calculators, right in the conversation. The announcement is on OpenAI's site.
For a small business: If someone on your team already uses ChatGPT, they will see this without doing anything. An answer that comes back as a small calculator or a chart could save a trip to a spreadsheet for quick questions. Anything that goes to a customer, or that involves a price or a commitment, still needs a person to check it first.
Google Cloud launches the Gemini agent for business
At Gemini at Work 2026 on October 8, Google Cloud announced the Gemini agent, which it describes as a universal agent for work. Google says it can plan tasks, use skills and tools, connect to a company's business systems, and return finished work inside documents, email and developer tools.
Google also says it picks a model for each job and includes cost controls plus the security, administration and governance features enterprises require. Details are in Google's post.
For a small business: The pitch is aimed at larger companies, so check what it would mean for an office of ten before getting excited. If your team already lives in Google's documents and email, it is worth watching. Letting an agent connect to your business systems is a decision about access, and an owner should make it deliberately.
Meta releases Muse Wonder and Muse Wonder Full
On October 12, Meta introduced Muse Wonder and Muse Wonder Full, which it calls its most capable models, aimed at personal and professional agent tasks. Muse Wonder is available through the Meta Model API and Muse Code. Muse Wonder Full is available only to Muse Code subscribers on the Power Usage plan or a new Maximum Usage plan. Meta published its pricing in the announcement.
Meta says Muse Wonder will later power features in its Muse app, starting with subscribers.
For a small business: This is mostly for people who build software. Unless you have a developer on staff, there is nothing to do yet. If Muse app features reach you later, treat them like any new tool and try them on a low-stakes task first.
Mistral opens a public preview of Mistral Large 4
On October 6, Mistral launched a public preview of Mistral Large 4 through the preview API in Mistral Studio. Mistral describes it as a 1 trillion-parameter multimodal model with 52 billion active parameters, and its largest model to date. The company says it will release the open weights by the end of October, along with more details on the architecture and training. See Mistral's announcement.
For a small business: Open weights matter to shops that want AI running on their own hardware so customer records stay in-house. That is a promise for later this month, not something to plan around today. Wait for the release and the details.
What this means for a New Hampshire shop
The common thread is that assistants are moving from answering to producing: a form, a chart, a finished document. That helps with recurring work such as quotes, schedules and follow-up emails. It also means more places where a draft can go out with a mistake in it.
Pick one recurring job and let an assistant draft it while a person approves every output. If you want help choosing the job, our team training is built around that kind of narrow start. Before buying anything, map one job first so you know what you are asking a tool to do.
If the news about open models makes you curious about keeping your data in-house, we can talk through what that involves. book a planning call for a free 30 minutes.