Today in DeAI: Latham & Watkins builds its own AI stack on open-weight models, DeepSeek keeps V4 Pro online past its shutdown date, and Morgan & Morgan pledges $1 billion to legal tech over a decade.
Latham & Watkins builds its own open-weight AI stack
Latham & Watkins, the second-largest US law firm by revenue at roughly $8.3 billion, has bought its own Nvidia GPU servers and is fine-tuning Nvidia's open-weight Nemotron 3 models on the firm's proprietary legal data for internal use, per the Financial Times. The drivers are keeping sensitive client data off external clouds and gaining flexibility versus buying consumption from frontier labs. Why it matters: this is the private-inference question answered at enterprise scale — when the data cannot leave the building, an $8.3 billion firm chose to become its own provider rather than rent a closed model. (Financial Times, Pulse 2.0) — our coverage
DeepSeek keeps V4 Pro online, legacy Flash names retire into V4.1
DeepSeek's pricing page confirms it is "continu[ing] to provide API services for DeepSeek V4 Pro after September 14, 2026, with the billing method remaining unchanged," reversing the planned phase-out. The same page notes the legacy deepseek-v4-flash model names are now retired and their requests are served by DeepSeek-V4.1-Flash at Flash pricing (cache-hit input from $0.003/M off-peak). Why it matters: builders on V4 Pro keep their endpoint and price, but anyone still calling the old Flash names is now silently on V4.1-Flash — a checkpoint change worth pinning in your config. (DeepSeek)
Morgan & Morgan pledges $1 billion to legal tech over 10 years
US personal-injury firm Morgan & Morgan committed to spend at least $1 billion on legal technology and AI over the next decade, unveiling its proprietary AI platform MX2, per Legal IT Insider. Why it matters: two large law firms in one week — Latham building in-house, Morgan & Morgan spending nine figures — signal that legal, a data-sensitive industry, is moving AI from a vendor subscription to a core internal capability. (Legal IT Insider)
Watching tomorrow
Whether Latham's build draws follow-on announcements from other large professional-services firms — and whether DeepSeek publishes a formal changelog entry for the V4 Pro reprieve beyond the pricing-page footnote.
Sources
- Latham & Watkins buys Nvidia GPU servers to set up in-house AI — Financial Times
- Latham & Watkins Buys Nvidia GPU Servers To Build In-House AI Systems — Pulse 2.0
- DeepSeek API Models & Pricing — DeepSeek
- Morgan & Morgan pledges $1bn tech spend over 10 years — Legal IT Insider
About DeAI
DeAI is an independent publication covering open-weight AI models, private inference, and decentralized infrastructure — the tools for running AI you actually control. We test providers on price, privacy, and refusal behavior and publish the numbers, not the vibes. DeAI is powered by Morpheus (mor.org), a decentralized inference marketplace, and covers it on the same terms as every other provider.
Powered by Morpheus and StrandCMS
Morpheus is a decentralized inference marketplace, covered on the same terms as every other provider — we rank it wherever the data lands. StrandCMS is the open-source, agent-first framework this site is built on.