The release includes DeepSeek-V4-Pro (1.6T total / 49B active) and DeepSeek-V4-Flash (284B total / 13B active), both trained natively at 1M context length.

Closed harnesses behind proprietary APIs force yielding control of agent memory to third parties.

Within a few months, they have more than a dozen production enterprise deployments & are processing over a billion events per hour.

Maine advances first statewide moratorium blocking data centers requiring over 20 megawatts

the most interesting detail here is how SkillClaw clusters cross-user trajectories into referenced skills and then uses the evolver to translate those patterns into concrete updates.

We test for a trend over time by fitting a weighted linear model to the log-odds of usage. Under this specification, Claude is the only AI service in the survey to show a statistically significant upward trend over this period

The ChatGPT for Excel add-in operates separately from your ChatGPT chat history. Conversations and data in Excel aren't shared with your ChatGPT chats, and activity doesn't sync between experiences at this time.

By default, data shared with ChatGPT isn't used to improve our models for ChatGPT Business, ChatGPT Enterprise, ChatGPT Edu, and ChatGPT for Teachers.

we collaborated with over 1,000 physicians to curate training data that enables more factual and comprehensive responses.

Support teams are high volume and high turnover, and thus need to train new reps in a fast and standardized way. To do so, they have clearly articulated standard operating procedures (SOPs) that guide the work of each rep. These SOPs create clear rules and guidelines that AI agents can model themselves off of.

MiniMax may have been able to get 100 billion tokens of data from interactions with Claude.

Legitimate Interest may be used for marketing purposes as long as it has a minimal impact on a data subject’s privacy and it is likely the data subject will not object to the processing or be surprised by it.