THE WORLD IS NOT STANDING STILL.RSS
BIG CHANGE.
BIG CHANGE, one tap away
BIG CHANGE, one tap away

Make room for the big picture.

Add BIG CHANGE to your home screen or desktop. It opens in its own window, with the latest stories when you’re online.

iPhone & iPad
Open bigchange.ai in Safari. Tap Share, then Add to Home Screen. Turn on Open as Web App if shown, then tap Add.
Android
Open bigchange.ai in Chrome. In the browser menu, choose Install app or Add to Home screen.
Desktop
In Chrome or Edge, use the install icon in the address bar or the browser’s app menu. In Safari on Mac, choose File → Add to Dock.
All briefings

Follow the change ·

The day,
in perspective.

What happened in AI and robotics — and the change behind it. A reading path through 14 stories from BIG CHANGE.

Ordered using RADAR attention, story opens and likes. Excerpts from our published reporting; open a story for the full context and sources. Dates in UTC.

guide

How to run a local forecast with Mistral Timestral

Mistral's Timestral package produces 21 quantiles per forecast step. This documentation-only guide covers local setup, output axes, covariates and a held-out comparison.

The big change

What changed: Mistral's October 15 release makes Timestral checkpoints available through a standalone Python forecaster. A local call takes numeric history, optionally conditions on past-only or genuinely known-future data, and returns 21 quantiles per future step.

Why it matters: An analyst can inspect forecast ranges for several related variables on local hardware. This provides a way to evaluate probabilistic forecasts without sending the history to a hosted inference API. The documented shape check shows how to read the output; it does not establish accuracy.

What to watch: The decision to use Timestral for a particular series turns on repeated time-ordered holdouts. Compare median error and interval coverage with an appropriate baseline across several cutoffs, using only covariates that were known at each cutoff. Those results determine whether its quantiles help with the task.

Read the full story
—
Back to the briefing
news

What two-channel audio adds to Xiaomi's MiDashengLM-Spatial

Xiaomi researchers released an audio-language model for spatial questions. Its preprint reports stronger directional benchmark scores with two-channel input, while complex scenes remain difficult.

The big change

What changed: The researchers added a spatial audio branch to MiDashengLM and released a paper, code and model checkpoint. The model takes binaural audio for spatial questions and retains a path for general audio understanding.

Why it matters: A text answer about a sound's location needs both event recognition and location cues. The dual-branch design lets the spatial branch use intermediate semantic features while leaving the semantic branch's forward path structurally unchanged.

What to watch: The authors' channel-swap benchmark tests whether left-right answers follow the audio channels. More independent tests, including varied recording setups and crowded scenes, are needed before treating the reported results as dependable in applications.

Read the full story
—
Back to the briefing
analysis

Microsoft’s MAI-Base-2 report measures pretraining, not a released model

Microsoft reports MAI-Base-2's held-out text results and fitted training-efficiency estimates. The base model is still in post-training, with no public checkpoint to test.

The big change

What changed: Microsoft has reported a completed large-scale base-model run, with held-out text evaluations and a detailed account of its training recipe. MAI-Base-2 remains in post-training.

Why it matters: Engineers can examine how Microsoft relates model quality to training compute, and which data, architecture and infrastructure choices it says contributed. The evidence concerns pretraining loss and Microsoft's internal measurements; there is no public MAI-Base-2 inference service or released checkpoint to test against a workload.

What to watch: Post-training results and access terms will determine whether this foundation becomes useful in products or to external teams. Microsoft's stated plan to license weights through Foundry is a future offering, not present availability.

Read the full story
—
Back to the briefing
news

NVIDIA releases Nemotron 3.5 Super VL weights for multimodal agents

NVIDIA released Nemotron 3.5 Super VL vision-language weights on October 15. The model supports text, images, video and tool-use workflows, with substantial hardware and evaluation requirements.

The big change

What changed: NVIDIA has released a post-trained model that combines visual input, text reasoning and tool use in one set of weights. Its documented output is text, so an application still supplies the tools and decides what the model may do.

Why it matters: A developer building document or video workflows can evaluate one model across visual evidence and subsequent tool calls, instead of treating visual interpretation as an isolated step. The benefit depends on results with the developer's own material and permissions.

What to watch: The released recipes make evaluation possible, but NVIDIA's training guide says some post-training data and components remain unavailable. Independent, task-specific results will say more about the model's usefulness than launch comparisons alone.

Read the full story
—
Back to the briefing
guide

Hy Memory 2.0 gives agents three ways to keep context

Tencent's Hy Memory 2.0.1 SDK separates chat, work and coding recall. Its MemHub app has source but no public release artifacts; integrations still need model endpoints and explicit access boundaries.

The big change

What changed: The published SDK exposes separate chat, working and coding memory paths through add() and search(), with source and workspace filters that an integrating application can use.

How it works: Chat organizes conversations into memory layers. Working extracts a session fact and, when warranted, reusable experiences from a tool trace. Coding adds a version chain for code changes. The caller supplies the source material, identities, storage configuration and model services.

Why it matters: An agent host can retrieve a compact record of prior conversations or task decisions instead of passing an entire history into every new session. Whether that record is accurate, authorized and useful still depends on what the host sends and which memories it allows back into the prompt.

Read the full story
—
Back to the briefing
news

Google's News AI pilot now includes more than 400 publications

Google expanded its paid News AI pilot to more than 400 publications. Its article overviews and audio briefings remain experiments without disclosed deal prices or traffic results.

The big change

What changed: Google reports that its News AI pilot has grown to more than 400 publications. It named new national and local partners on October 15, nearly a year after announcing the pilot.

Why it matters: Participating publishers receive payment for supplying content to specific AI news experiments. Readers may encounter an overview or audio briefing before visiting the reporting outlet. The public announcements do not show how those features affect visits or subscriptions.

What to watch: Google has not disclosed comparable contract terms or measured outcomes across the partners. Those details would let publishers judge whether the licensing income and any resulting audience engagement justify participation.

Read the full story
—
Back to the briefing
news

Attie adds prompt-built social pages for Bluesky communities

Attie says current users can make small social sites with Bluesky participation. Its public sign-in flow still labels the service early access and includes a waitlist option.

The big change

What changed: Attie began with custom social feeds. Its new builder lets people who already have Attie access describe a small site in plain words and share the resulting page. The announcement covers current Attie users; the public site still labels the service early access.

Why it matters: A poll or AMA can now sit at its own address while inviting people from an existing Bluesky community to participate through their social identity. For an organizer, that connects a group activity to the audience already on Bluesky. Attie presents Bluesky sign-in as the participation path.

What to watch: Creator access and the controls behind a real community activity remain the practical decisions. Attie's public sign-in flow includes a waitlist option, while the sources do not establish which new accounts can create immediately or document moderation controls, usage limits or reliability.

Read the full story
—
Back to the briefing
news

Google's Universal Cart rolls out in the U.S. across Search and Gemini

Google's shared cart now rolls out in U.S. Search and Gemini. It can track products from several retailers, while checkout follows Google Pay or each merchant's own cart and the merchant remains the seller.

The big change

What changed: A cart that Google previewed in May is now rolling out across Search and Gemini in the U.S. Shoppers can bring finds from several retailers into one saved list, while each purchase still belongs to its merchant.

Why it matters: The list puts saved items, price and stock notices, promotions and Wallet card perks near the purchase decision. Google gains a role in organizing the shopping trip, but the retailer remains the seller and handles confirmation and support.

What to watch: At checkout, the route offered for each item matters: Google Pay in the cart or a transfer to the merchant's cart. Shoppers need to check the final terms with the seller. Google's launch post gives no complete list of eligible products or retailers.

Read the full story
—
Back to the briefing