#219 - GPT 5, Opus 4.1, OpenAI's Open Source, Astrocade | Last Week in AI

In this episode of the Last Week in AI podcast, Andrey Kurenkov and Jeremie Harris discuss the latest AI news, including the release of GPT-5, Opus 4.1, and Gemini DeepThink AI. They analyze the capabilities, benchmarks, and potential impacts of these models, as well as business updates from Meta, Microsoft, OpenAI, and Anthropic, including revenue, funding rounds, and market share. The hosts also cover open-source projects like OpenAI's GPT-OSS and Falcon H1, research advancements such as Google's Alpha Earth Foundations and Genie3, and policy and safety concerns related to AI, including export bans and model alignment.

Outlines

Part 1: AI Landscape & GPT-5

Part 2: Model Updates & Business

Part 3: Open Source Models & Architectures

Part 4: Safety, Alignment & Global AI

Sign in to continue reading, translating and more.

Open full episode in Podwise

#219 - GPT 5, Opus 4.1, OpenAI's Open Source, Astrocade

Last Week in AI

Part 1: AI Landscape & GPT-5

Introduction to the Current State of AI and GPT-5

GPT-5 Details: System Architecture, Evals, and User Experience

GPT-5 Performance Metrics, Pricing, and the M-Dash Quirk

Part 2: Model Updates & Business

Opus 4.1, Gemini DeepThink AI, and Grok Imagine's Uncensored Content

Meta and Microsoft's AI Investments, OpenAI's Stargate Norway Data Center

Anthropic's Revenue and Fundraising, Enterprise LLM Market Share

OpenAI's Revenue, Fundraising, and Noma Security's Focus on AI Security

Part 3: Open Source Models & Architectures

OpenAI's Open Source Model Release: GPT-OSS

GPT-OSS Technical Details, Usage, and Comparison to Other Models

Falcon H1: A Hybrid Head Language Model

MetaClip-2: Worldwide Scaling Recipe

Flux.1 Craya and Google's AI Model for Climate Change Tracking

Google's Genie3 and Model Architecture Discovery

Concerns About Information Leakage in Model Architecture Discovery

Part 4: Safety, Alignment & Global AI

Meter Evaluation of Grok 4 and OpenAI's Risk Assessment of Open Weight LLMs

Anthropic's AI Vaccine and Alignment Auditing Agents

Obfuscating Chains of Thought and China's Compute Capabilities

China's AI Agenda and Global Governance Efforts

Export Bans and Listener Comments on Chatbot Bias

Addressing Listener Feedback on Chatbot Bias and Grok

#219 - GPT 5, Opus 4.1, OpenAI's Open Source, Astrocade

Last Week in AI

Part 1: AI Landscape & GPT-5

00:11Introduction to the Current State of AI and GPT-5

Introduction to the Current State of AI and GPT-5

05:04GPT-5 Details: System Architecture, Evals, and User Experience

GPT-5 Details: System Architecture, Evals, and User Experience

14:22GPT-5 Performance Metrics, Pricing, and the M-Dash Quirk

GPT-5 Performance Metrics, Pricing, and the M-Dash Quirk

Part 2: Model Updates & Business

21:12Opus 4.1, Gemini DeepThink AI, and Grok Imagine's Uncensored Content

Opus 4.1, Gemini DeepThink AI, and Grok Imagine's Uncensored Content

25:50Meta and Microsoft's AI Investments, OpenAI's Stargate Norway Data Center

Meta and Microsoft's AI Investments, OpenAI's Stargate Norway Data Center

32:12Anthropic's Revenue and Fundraising, Enterprise LLM Market Share

Anthropic's Revenue and Fundraising, Enterprise LLM Market Share

37:12OpenAI's Revenue, Fundraising, and Noma Security's Focus on AI Security

OpenAI's Revenue, Fundraising, and Noma Security's Focus on AI Security

Part 3: Open Source Models & Architectures

42:16OpenAI's Open Source Model Release: GPT-OSS

OpenAI's Open Source Model Release: GPT-OSS

48:07GPT-OSS Technical Details, Usage, and Comparison to Other Models

GPT-OSS Technical Details, Usage, and Comparison to Other Models

53:30Falcon H1: A Hybrid Head Language Model

Falcon H1: A Hybrid Head Language Model

57:39MetaClip-2: Worldwide Scaling Recipe

MetaClip-2: Worldwide Scaling Recipe

1:01:12Flux.1 Craya and Google's AI Model for Climate Change Tracking

Flux.1 Craya and Google's AI Model for Climate Change Tracking

1:04:50Google's Genie3 and Model Architecture Discovery

Google's Genie3 and Model Architecture Discovery

1:13:37Concerns About Information Leakage in Model Architecture Discovery

Concerns About Information Leakage in Model Architecture Discovery

Part 4: Safety, Alignment & Global AI

1:17:22Meter Evaluation of Grok 4 and OpenAI's Risk Assessment of Open Weight LLMs

Meter Evaluation of Grok 4 and OpenAI's Risk Assessment of Open Weight LLMs

1:23:12Anthropic's AI Vaccine and Alignment Auditing Agents

Anthropic's AI Vaccine and Alignment Auditing Agents

1:27:11Obfuscating Chains of Thought and China's Compute Capabilities

Obfuscating Chains of Thought and China's Compute Capabilities

1:31:25China's AI Agenda and Global Governance Efforts

China's AI Agenda and Global Governance Efforts

1:38:11Export Bans and Listener Comments on Chatbot Bias

Export Bans and Listener Comments on Chatbot Bias

1:42:35Addressing Listener Feedback on Chatbot Bias and Grok

Addressing Listener Feedback on Chatbot Bias and Grok