What you'll learn

• 1 (00:02) **Why Inkling Matters - The Thesis** - Jordan introduces Inkling, a new open-source model from Thinking Machines Lab (led by former OpenAI CTO Mira Marathi), as potentially the most important AI model most people will never use because it signals a strategic shift in enterprise AI procurement.
• 2 (03:43) **Setting the Stage: Why Frontier Models Are Too Much for Most Enterprises** - Jordan explains that for many companies outside the Fortune 500, frontier models are overkill and too expensive, making open-source fine-tuning a newly relevant option.
• 3 (05:56) **Inkling Model Specs and Capabilities** - The host reads directly from Thinking Machines' release, detailing Inkling's architecture and positioning.
• 4 (08:14) **Benchmarks: Inkling vs. the Frontier** - Jordan shows the Artificial Analysis Intelligence Index, placing Inkling at 41 compared to leaders like Fable 5 (60) and GPT-5.6 Sol (59).
• 5 (12:42) **The Real Product: Tinker Fine-Tuning as a Service** - Jordan explains that Inkling's massive size (600GB compressed) means most companies can't run it locally, so Thinking Machines' business model is Tinker, which manages training without requiring customer-owned GPUs.
• 6 (13:22) **Strategic Reset #1: American Open Weights Reopen Procurement** - Many US enterprises couldn't use Chinese open-source models due to geopolitical risks, government contracts, and trust concerns.
• 7 (15:52) **Strategic Reset #2: Model Overhang Makes Customization Timely** - Jordan connects the "model overhang" (frontier models far exceeding enterprise needs) to the new viability of fine-tuning.

+ Full timestamped outline available in the app