- H-FARM AI's Newsletter
- Posts
- Claude now reads your bank accounts
Claude now reads your bank accounts
PLUS: Google ships Gemini Omni 1.2 Flash, cuts prices 30% & OpenAI ships GPT-6 Cyber to find exploits. NVIDIA VFlash predicts robot motion in parallel, ChatGPT search hits 20% of Google's.

1️⃣ Anthropic lets Claude users link bank and card accounts to query their spending, in beta for US paid plans. 2️⃣ Google ships Gemini Omni 1.2 Flash with 15-second clips and API pricing cut 30% to $0.07 per second. 3️⃣ OpenAI releases GPT-6 Cyber, a model that finds and validates exploitable vulnerabilities. |
|
MAIN AI UPDATES / 6th November 2026
💳️ Claude now reads your bank accounts 💳️
Bank account integration arrives in Claude, starting in beta.
Anthropic has launched Personal Finance in Claude, letting users link bank and card accounts and ask plain-language questions about balances, transactions and spending. Think "how much did I spend on restaurants last month?" answered in seconds, no spreadsheets required. The feature is rolling out in beta to US Pro, Max and Ultra plans, with a clear limit: Claude cannot move or invest money. It is a strong adoption signal that assistants are entering everyday finance, a space owned by banking and budgeting apps. Trust and data privacy will likely decide how quickly users connect their accounts.
🎬️ Google ships Gemini Omni 1.2 Flash, cuts prices 30% 🎬️
Longer, steadier AI video clips at a lower API price for developers.
Google has shipped Gemini Omni 1.2 Flash, the latest version of its video generation model, with two upgrades developers will notice right away. The first is quality. Clips now run up to 15 seconds, cinematic controls have been improved for finer shot direction, and the model reduces the visual drift that often makes AI-generated characters and scenes morph mid-clip. The second, and arguably more important, is price. API pricing drops 30% to $0.07 per second for 720p output, a cut that puts direct pressure on rival video APIs. Longer, cheaper and more stable clips move the model beyond experiments and make it a practical tool for ads, social content and product demos.
🛡️ OpenAI ships GPT-6 Cyber to find exploits 🛡️
A dedicated model bringing speed to vulnerability research.
OpenAI has released GPT-6 Cyber, a dedicated cybersecurity model built to identify and validate exploitable vulnerabilities. The key detail is control: the model works within boundaries defined by researchers, keeping testing limited to approved targets and scopes. Validation is what makes it useful, since security teams lose hours triaging findings that turn out to be harmless; confirming whether a flaw is actually exploitable can shorten that work. It is a capability jump that cuts both ways for defenders and attackers, which is why the guardrails sit at the center of the launch. GPT-6 Cyber shows specialized models moving into core security workflows.
INTERESTING TO KNOW
🦾 NVIDIA VFlash predicts robot motion in parallel 🦾
NVIDIA VFlash is a new vision model built for speed: instead of generating outputs token by token, it predicts geometry such as robot trajectories and 3D motion in parallel. That matters for real-time robotics, where latency decides whether a machine reacts in time, adding competitive pressure on slower sequential approaches.
🔎 ChatGPT search hits 20% of Google's 🔎
OpenAI says ChatGPT search volume is now roughly 20% of Google's, showing chat is becoming a partial replacement for the classic search box. The assistant counts 1.2 billion weekly users, and 250 million of them use voice. For brands and publishers, it is a clear adoption signal: visibility inside AI answers now matters alongside traditional SEO rankings.

📩 Have questions or feedback? Just reply to this email , we’d love to hear from you!
🔗 Stay connected:
