#234 - Opus 4.6, GPT-5.3-codex, Seedance 2.0, GLM-5 | Last Week in AI

The Last Week in AI podcast delivers a dense recap of recent AI advancements, focusing on groundbreaking model releases. Opus 4.6, Codex 5.3, and Gemini 3 DeepThink are highlighted for their enhanced capabilities and potential economic impacts, particularly in white-collar markets. The discussion covers the shift towards AI as a universal knowledge worker, improvements in coding-oriented models, and the challenge of evaluating AI safety and security. The hosts also explore ByteDance's Seedance 2.0 video model and the implications of relaxed copyright policies in China. Additionally, they touch on AI company valuations, funding rounds for robotics firms like Aptronic, and the ongoing AI talent war, including leadership changes at XAI.

Outlines

Part 1: Introduction, Listener Engagement

Part 2: Model Releases, Capabilities, Benchmarks

Part 3: Business, Funding, Industry Shifts

Part 4: Research, Alignment, Safety

Part 5: Conclusion

Sign in to continue reading, translating and more.

Open full episode in Podwise

#234 - Opus 4.6, GPT-5.3-codex, Seedance 2.0, GLM-5

Last Week in AI

Part 1: Introduction, Listener Engagement

Podcast Introduction, Listener Feedback, and Focus on AI Content

Part 2: Model Releases, Capabilities, Benchmarks

Anthropic's Opus 4.6 and OpenAI's Codex: New Model Releases and Capabilities

GPT-5.3 Codex Details, Recursive Self-Improvement, and Cerebrus Partnership

Google's Gemini 3 DeepThink, Seedance 2.0, and AI Video Generation

GLM-5, DeepSeek, Cursor Composer 1.5, and XAI's Grok Imagine API

Part 3: Business, Funding, Industry Shifts

11Labs and Runway Raises, Valuations, and AI Business Landscape

XAI's Reorganization, Waymo's Hardware, and Open Source Projects

Part 4: Research, Alignment, Safety

Research Advancements: Learning to Reason in 13 Parameters and RL for LLM Agents

Opus 4.6 on VendingBench, Metter Evals, and Anthropic's Alignment Research

Part 5: Conclusion

Podcast Outro and Musical Interlude

#234 - Opus 4.6, GPT-5.3-codex, Seedance 2.0, GLM-5

Last Week in AI

Part 1: Introduction, Listener Engagement

00:11Podcast Introduction, Listener Feedback, and Focus on AI Content

Podcast Introduction, Listener Feedback, and Focus on AI Content

Part 2: Model Releases, Capabilities, Benchmarks

04:42Anthropic's Opus 4.6 and OpenAI's Codex: New Model Releases and Capabilities

Anthropic's Opus 4.6 and OpenAI's Codex: New Model Releases and Capabilities

14:10GPT-5.3 Codex Details, Recursive Self-Improvement, and Cerebrus Partnership

GPT-5.3 Codex Details, Recursive Self-Improvement, and Cerebrus Partnership

23:53Google's Gemini 3 DeepThink, Seedance 2.0, and AI Video Generation

Google's Gemini 3 DeepThink, Seedance 2.0, and AI Video Generation

33:20GLM-5, DeepSeek, Cursor Composer 1.5, and XAI's Grok Imagine API

GLM-5, DeepSeek, Cursor Composer 1.5, and XAI's Grok Imagine API

Part 3: Business, Funding, Industry Shifts

42:5311Labs and Runway Raises, Valuations, and AI Business Landscape

11Labs and Runway Raises, Valuations, and AI Business Landscape

57:34XAI's Reorganization, Waymo's Hardware, and Open Source Projects

XAI's Reorganization, Waymo's Hardware, and Open Source Projects

Part 4: Research, Alignment, Safety

1:07:55Research Advancements: Learning to Reason in 13 Parameters and RL for LLM Agents

Research Advancements: Learning to Reason in 13 Parameters and RL for LLM Agents

1:17:12Opus 4.6 on VendingBench, Metter Evals, and Anthropic's Alignment Research

Opus 4.6 on VendingBench, Metter Evals, and Anthropic's Alignment Research

Part 5: Conclusion

1:28:30Podcast Outro and Musical Interlude

Podcast Outro and Musical Interlude