Episode
Anthropic launches Claude Science and a new model
- Podcast
- Last Week in AI
- Published
- Jun 30, 2026
- Duration seconds
- 1014
- Processing state
processed
Actions
POST https://stenobird.com/v1/public/podcasts/last-week-in-ai-6751488/episodes/anthropic-launches-claude-science-and-a-new-model/transcription-requests
Idempotently request low-priority transcript generation for this episode.GET https://stenobird.com/podcast/last-week-in-ai-6751488/anthropic-launches-claude-science-and-a-new-model.md
Read the agent-friendly Markdown representation of this episode resource.
Summary
Anthropic is pivoting toward specialized scientific workflows with Claude Science, while XAI opens its ecosystem via a new MCP server. The episode also covers Google's aggressive optimization of Gemini models for speed and cost-efficiency.
Topics
- Anthropic
- Claude Science
- XAI
- Google Gemini
- Model Context Protocol
- AI Agents
- Machine Learning
- Claude Sonnet
Highlights
- Main idea: Anthropic is prioritizing specialized research workflows over raw model capacity with the launch of Claude Science
- Main idea: XAI's new MCP server allows developers to integrate real-time X data directly into tools like Claude and Cursor
- Practical takeaway: Google's Gemini Omni Flash offers high-speed, multimodal video generation at a competitive cost per second
- Failure mode: Rapidly changing models, like the transition to Nano Banana 2 Lite, can break existing prompt engineering and workflows
- Market trend: The industry is shifting toward agentic coding and highly efficient, low-cost models like Claude Sonnet 5
Chapters
1:00Anthropic's Claude Science Launch: An overview of the new Claude Science workbench, connecting 60 scientific databases for research automation.4:00The Rise of Base 44: Analyzing the rapid growth and $100M ARR milestone of the coding-focused platform Base 44.7:00XAI and the New MCP Server: How X's new Model Context Protocol allows Claude and Cursor to tap into real-time X account data and sentiment.13:00Efficiency in the Model Tier: Discussing the shift toward cost-per-task efficiency and the performance of newer, smaller models.14:00Google's Gemini Optimizations: A look at Gemini Omni Flash and Nano Banana 2 Lite, focusing on speed, cost, and multimodal capabilities.