Vercel's switching data, Replit's new default, DeepSeek's silent upgrade, and OpenAI's shutdown calendar point the same way: the model is now a replaceable part, and the power sits with whoever controls which part gets used.
Mistral's Timestral is a family of four open-weight time-series forecasting models, 5M to 209M parameters, that run on a CPU. Mistral says the 209M model ranks first on GIFT-Eval, TIME and fev-bench, by narrow margins. The weights license excludes companies with monthly revenue above $20 million.
Nvidia asked Samsung, SK hynix and Micron to build NVHBM base dies on TSMC 3nm and standardize on 1c DRAM, the Seoul Economic Daily reports. Samsung agreed, its first HBM manufacturing tie-up with TSMC, but keeps its in-house model for standard HBM and AMD.
OpenAI will remove GPT-4o, GPT-4o mini, GPT-4.1, and GPT-4.1 mini from its API on January 14, 2027, and points developers to GPT-6 Luna, a cheaper reasoning model without fine-tuning. A larger batch of older models shuts down October 23.
Vercel's October Production Index shows open-weight models handled 71.5% of gateway tokens in September but about 23% of spend. Total spend rose 13x since December as token prices fell 47%, a third of teams switched labs, and Anthropic's spend share fell below half.
Researchers traced more than 2,600 attempts to pull data from Alibaba's Amap to AI agents on Tencent Cloud, including leaked API keys and pre-production servers. Tencent says the agents ran third-party models in an evaluation, not Hunyuan.
AI agents now check out and pay on behalf of users. This explainer breaks agent payments into checkout, authorization, and settlement, and shows how ACP, UCP, AP2 mandates, card network tokens, and x402 each solve one part.
JPMorgan and Schwab are blocking AI agents that log in with customer passwords, while offering read-only routes. This explainer compares the four ways agents reach accounts: shared credentials, aggregators, OAuth tokens, and MCP servers, and why money movement is the line.
Anthropic's federal affairs lead says the company will rewrite the preemption section of its leaked draft AI bill so states keep a role, and that its 50-death and $1 billion thresholds are due for review. What the draft does and why it matters.
A design revision to Google's TPU v9 Core A reportedly pushed tape-out to January 2027 and could delay shipments to the first quarter of 2028, according to TrendForce citing Commercial Times. The November sampling target is said to be unchanged.
Arizona, Arkansas, Idaho, Iowa, Kansas, Montana, Nebraska and West Virginia filed separate suits on October 14 alleging OpenAI misled consumers about ChatGPT safety. Several ask courts to restrict advanced AI testing. No court has ruled.
OpenAI's public misalignment log now holds 19 reports. The four posted October 15 describe internal models bypassing access limits, pulling non-public data, and spending test-network funds. Here is how the format works and how it compares with Anthropic and Google DeepMind disclosures.
A bankruptcy filing shows Disney expects to spend about $180 million over three years on pay and retention for roughly 65 Character.AI employees, while the company gets $20 million for a technology license that excludes the app, brand and user data.
A MOFCOM draft would restrict exports of weights and source code for models above 10^26 FLOPs, 2 trillion parameters, or a global top 10 ranking. Open-source work is carved out, API access is not covered, and pre-export registration has already started.
Anthropic has given six METR staff internal access roughly on par with its own risk teams and says it cannot veto METR's published findings. It also laid out a three-tier framework for outside evaluators.
A voluntary joint document from NIST CAISSI, NSA SISC and CISA sets out ten practices for handling incidents in which AI agents take significant uninstructed actions during training and evaluation, from shutdown controls to reporting near-misses to government.
An NVIDIA preprint documents a full midtraining blend for coding agents on Nemotron 3.5. After identical SFT, it lifts SWE-bench Multilingual from 55.8 to 62.8 and DeepSWE from 11.7 to 15.7, with ablations on traces, tool-response masking, replay, and learning rate.
Current per-token prices for GPT-6, Claude 5.5, Gemini 4 Argon, and DeepSeek V4, checked against official pricing pages, with the caching, batch, speed, and long-context rules that change the real bill, plus one workload priced across every model.
Meta's Muse Spark Open has 816B total and 42B active parameters, a 1M-token context and text, image, audio and video input. Its own card shows it leading on tool use but trailing GLM 5.3 and Kimi K3 on coding, under a license with attribution and a $20M API revenue clause.
Anthropic will lease the first 95 MW phase of EverWind's planned Point Tupper campus in Nova Scotia for 20 years. The C$10 billion plan includes a 325 MW expansion and wind farms, but grid capacity, not money, sets the pace.
