Podcast #18

The AI landscape is shifting rapidly with major developments across pricing, hardware, and corporate rivalries. OpenAI is reportedly preparing to launch a new subscription tier called 'ChatGPT Pro Lite', priced at $100 per month. This plan targets heavy users who frequently hit the rate limits of the current Plus tier but don't require the full unlimited scope of the Pro version.

In a dramatic turn of events, Anthropic has accused competitors like DeepSeek, Moonshot AI, and MiniMax of 'distillation' attacks. They claim these companies created over 24,000 fake accounts to generate millions of interactions with Claude, effectively trying to clone its reasoning and coding capabilities.

On the technical front, OpenAI's GPT-5 Codex has demonstrated immense endurance and capability by running a 25-hour continuous coding sprint. In this stress test, the model autonomously built a design tool using over 13 million tokens and producing 30,000 lines of code, showcasing a leap in agentic behavior.

Hardware is also seeing massive gains with Taalas revealing their HC1 chip. This new silicon hardwires Large Language Models directly into the chip, achieving a staggering inference speed of 17,000 tokens per second per user. This breakthrough promises to make AI interactions instantaneously responsive.

Lastly, the corporate world is feeling the heat. IBM's stock took a hit after Anthropic highlighted that their Claude Code tools can rapidly refactor old COBOL applications, signaling a major disruption for legacy tech maintenance. Meanwhile, OpenAI is cementing its enterprise dominance by securing multiyear deals with consulting giants like Accenture and McKinsey to deploy their 'Frontier' platform.