Subvolts take Track GPT-5.5, Grok Voice, and DeepSeek V4 alongside major lab investments and safety research. May 3, 2026 - Developers
- Learn about GPT-5.5's coding enhancements and DeepSeek V4's large context window for building advanced applications.
- Business
- Understand the competitive landscape with new voice AI from xAI and major lab investments impacting the market.
- Everyday users
- Discover new AI voice capabilities and how models like GPT-5.5 are improving, with insights into AI safety research.
This episode covers significant new model releases and substantial industry investments, offering a broad overview of current AI progress.
#243 - GPT 5.5, DeepSeek V4, AI safety sabotage
Audio streams from Skynet Today. Your place is remembered on this device.
From Subvolts
Summary and analysis by Subvolts editorial, AI-assisted.
This episode of Last Week in AI features hosts Andrey Kurenkov and Jeremie Harris discussing OpenAI's GPT-5.5 release, highlighting coding improvements, higher pricing, and system prompt quirks. The show also covers xAI's Grok Voice Think Fast 1.0, DeepSeek's open-sourced V4 models with large context windows, and significant business and policy updates including Google's investment in Anthropic and new AI safety research.
This episode covers new AI models like GPT-5.5 and DeepSeek V4, which are better at tasks like coding and understanding long texts. It also discusses new AI voice technology and major company investments, explaining how these advancements impact businesses and users.
This episode offers a comprehensive AI news roundup for those tracking major lab developments and industry shifts. Listeners will learn about the latest model advancements from OpenAI and DeepSeek, understand the competitive landscape with xAI's new voice product, and grasp the implications of significant funding and policy news. It's essential for AI professionals, researchers, and enthusiasts staying informed on the rapid pace of AI innovation and its business integration.
- HostAndrey Kurenkov
- HostJeremie Harris
- Episode Title: #243 - GPT 5.5, DeepSeek V4, AI safety sabotage
- Show: Last Week in AI
- Length: 112 minutes
- GPT-5.5 features: Strong coding improvements, chain-of-thought monitorability, misalignment testing, higher pricing than GPT-5.4, system-prompt warning about 'goblins'.
- Grok Voice Think Fast 1.0 claims: Large benchmark leads for real-time voice agents, impact on Starlink customer support and sales conversion.
- DeepSeek V4 features: MoE scaling, 1M-token context via hybrid/compressed attention.
Our 243rd episode with a summary and discussion of last week's big AI news!
Recorded on 04/29/2026
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-5.5 with strong coding-oriented improvements, a system card discussing chain-of-thought monitorability and misalignment testing, higher pricing than GPT-5.4, and notable quirks like a system-prompt warning about “goblins.”
- xAI launched Grok Voice Think Fast 1.0, claiming large benchmark leads for real-time voice agents and reporting major Starlink customer-support automation and sales conversion impact.
- DeepSeek open-sourced DeepSeek V4 (Pro and Flash) featuring MoE scaling and 1M-token context via hybrid/compressed attention changes, while Tencent released Hunyuan 3 preview with weaker benchmark performance; a new long-horizon agent benchmark (Clawmark) shows low task success rates.
- Major business, legal, and policy updates include Google’s planned up-to-$40B investment and 5GW compute commitment to Anthropic, Meta’s AWS Gravitron deal and China blocking Meta’s Manus acquisition, a revamped OpenAI–Microsoft agreement, ongoing Musk–OpenAI trial developments, and new safety/security research on sabotage, document degradation under delegation, and bit-flip attacks.
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
- (00:02:00) News Preview
- (00:02:26) Response to listener comments
- (00:02:55) Sponsors
- Tools & Apps
- (00:05:55) OpenAI Unveils Its New, More Powerful GPT-5.5 Model - The New York Times
- (00:23:33) xAI Launches grok-voice-think-fast-1.0: Topping τ-voice Bench at 67.3%, Outperforming Gemini, GPT Realtime, and More - MarkTechPost
- (00:29:00) Claude can now plug directly into Photoshop, Blender, and Ableton | The Verge
- Projects & Open Source
- (00:29:38) China's DeepSeek releases preview of long-awaited V4 model as AI race intensifies
- (00:47:05) Tencent Unveils Hy3 preview; Model Enhances Agent Capabilities and Real-World Usability - Tencent 腾讯
- (00:50:14) Jargon decoded
- MoE
- Andrey Kurenkov
- Mixture of Experts, a technique to make AI models more efficient by using specialized parts for different tasks.
- chain-of-thought
- Andrey Kurenkov
- A method where AI explains its reasoning step-by-step, like showing its work in math.
- misalignment testing
- Andrey Kurenkov
- Checking if an AI's goals and actions match human intentions and safety guidelines.
2 h 09
57 min
53 minQuestions people ask
- What are the key features of OpenAI's GPT-5.5?
- GPT-5.5 offers strong coding-oriented improvements, higher pricing than GPT-5.4, and includes features like chain-of-thought monitorability and misalignment testing.
- What is DeepSeek V4 and what are its capabilities?
- DeepSeek V4 is an open-sourced model featuring MoE scaling and a 1 million token context window, achieved through hybrid and compressed attention mechanisms.
Page generated Oct 19, 2026. Summaries are Subvolts' own; the audio, artwork and notes belong to Skynet Today.
