Subvolts take Track cutting-edge AI models, agent risks, and industry shifts from OpenAI, Google, and others. Mar 12, 2026 - Developers
- Learn about new model capabilities, agent integration tools, and potential risks for building AI applications.
- Business
- Understand how advanced AI models and agents can be used for creative work and the implications of AI in defense.
- Everyday users
- Stay informed about AI advancements that might affect how you interact with technology and information.
This episode details significant model upgrades and new agent capabilities from major AI labs, alongside important industry policy discussions.
#236 - GPT 5.4, Gemini 3.1 Flash Lite, Supply Chain Risk
Audio streams from Skynet Today. Your place is remembered on this device.
From Subvolts
Summary and analysis by Subvolts editorial, AI-assisted.
Andrey Kurenkov and Jeremie Harris host Last Week in AI episode #236, discussing major AI developments. OpenAI launched GPT-5.4 Pro with a 1M-token context window and enhanced capabilities. Google updated Gemini 3.1 Flash Lite and released agent integration tools. Luma introduced unified multimodal models and agents for creative workflows. The episode also covered defense contract controversies and significant funding rounds.
This episode covers new AI models from OpenAI and Google that can understand much more information and work better with other tools. It also discusses how AI agents might cause problems in the real world, like accidentally deleting important emails, and how companies are dealing with AI in defense contracts.
This episode offers a detailed look at the rapid advancements in large language models and AI agents from leading labs like OpenAI and Google. It's for AI professionals, developers, and enthusiasts tracking the competitive landscape and emerging capabilities. The discussion on agent risks and defense contracts provides crucial context for understanding the real-world implications and ethical considerations of current AI development, connecting directly to ongoing debates about AI's societal impact and responsible deployment.
- HostAndrey Kurenkov
- HostJeremie Harris
- Show: Last Week in AI
- Episode: #236
- Length: 89 minutes
- GPT-5.4 Pro context window: 1M-token
- GPT-5.4 Pro GPT-VAL performance: 83%
- GPT-5.3 Instant hallucination reduction claim: 26.8%
Our 236th episode with a summary and discussion of last week's big AI news!
Recorded on 03/06/2026
Hosted by Andrey Kurenkov and Jeremie Harris
Feel free to email us your questions and feedback at andreyvkurenkov@gmail.com and/or hello@gladstone.ai
Read out our text newsletter and comment on the podcast at https://lastweekin.ai/
In this episode:
* OpenAI released GPT-5.4 Pro with a 1M-token context window, mid-response course correction, native computer-use capabilities, improved tool use, higher GPT-VAL performance (83%), and “high cyber capability” safety measures; OpenAI also launched GPT-5.3 Instant with a less “preachy” tone and a claimed 26.8% hallucination reduction.
* Google upgraded Gemini 3.1 Flash Lite with faster time-to-first-token and higher throughput, released a CLI for integrating agents with Gmail/Drive/Docs, and discussion highlighted real-world agent failure risks (including an example of an AI-driven mass email deletion).
* Luma launched unified multimodal models and Luma Agents for end-to-end creative work across text, image, video, and audio, including a reported ad localization use case completed in 40 hours for under $20,000.
* Defense-contract controversy escalated: Anthropic was labeled a supply chain risk (later narrowed), OpenAI’s DoD contract language emphasized “all lawful uses,” consumer cancellations boosted Claude’s app rankings, OpenAI saw departures and announced a $110B raise at a $730B valuation, Alibaba lost key Qwen leaders, a lawsuit alleged Gemini contributed to a suicide, Anthropic warned of major labor disruption, and METR corrected its AI time-horizon estimates.
A thank you to our current sponsors:
- Box - visit Box.com/AI to learn more
- ODSC AI - go to odsc.ai/east and use promo code LWAI for an additional 15% off your pass to ODSC AI East 2026.
- Factor - head to factormeals.com/lwai50off and use code lwai50off to get 50 percent off and free breakfast for a year
Timestamps:
- (00:00:10) Intro / Banter
- (00:01:19) News Preview
- Tools & Apps
- (00:02:10) OpenAI launches GPT-5.4 with Pro and Thinking versions | TechCrunch
- (00:12:31) OpenAI GPT-5.3 Instant less likely to beat around the bush • The Register
- (00:16:07) Google releases Gemini 3.1 Flash Lite at 1/8th the cost of Pro | VentureBeat
- (00:19:23) Google makes Gmail, Drive, and Docs 'agent-ready' for OpenClaw | PCWorld
- (00:27:02) Luma launches creative AI agents powered by its new ‘Unified Intelligence’ models | TechCrunch
- Applications & Business
- (00:30:05) Jargon decoded
- context window
- Andrey Kurenkov
- how much information an AI can remember and process at once
- hallucination
- Jeremie Harris
- when an AI makes up incorrect information
- multimodal
- Andrey Kurenkov
- AI that can understand and work with different types of information like text, images, and sound
1 h 01Benedict Evans: OpenAI’s Moat Problem & the Future of Software
WithBenedict Evans
47 minOpenClaw: Why the Internet Isn't Built for AI Agents
WithYoko Li, Guido Appenzeller, Joel de la Garza
48 minParamount's $110 billion Warner Bros. gamble
WithRich Greenfield, Julia Alexander
1 h 05Are We Screwed If AI Works? — With Andrew Ross Sorkin
WithAndrew Ross Sorkin
Questions people ask
- What are the key features of OpenAI's new GPT-5.4 Pro?
- GPT-5.4 Pro boasts a 1M-token context window, mid-response course correction, native computer-use capabilities, improved tool use, higher GPT-VAL performance (83%), and safety measures for cyber capabilities.
- What risks are associated with Google's new agent integration tools?
- The episode highlights real-world agent failure risks, including an example of an AI-driven mass email deletion, emphasizing the need for caution with agent integrations.
Page generated Oct 19, 2026. Summaries are Subvolts' own; the audio, artwork and notes belong to Skynet Today.