Real Alpha
Watch mode

All-In · 2 min · Oct 17, 2026

A small model cut Coinbase fraud by about 30%

David Sacks

Co-founder & partner, Craft Ventures

PayPal's first COO, founded Yammer; White House AI & crypto czar 2025–26.

  • Chamath PalihapitiyaHost · Founder & CEO, Social Capital

    Led Facebook's growth team; his firm backed Slack and AI chipmaker Groq.

  • Jason CalacanisHost · Founder & CEO, LAUNCH

    Early Uber angel; sold Weblogs Inc. to AOL; hosts This Week in Startups.

About this segment

Sacks cites Coinbase fine-tuning a 9-billion-parameter open model to detect fraud, cutting fraudulent transactions by about 30% in tests. The example suggests many enterprise tasks can use cheaper specialist models.

From “CoreWeave Crash, AI Trade Cracks, Anthropic IPO Red Flags, Bill Gates Doomer”, Oct 17, 2026. Watch the full episode

“In testing, it cut fraudulent transactions by about 30% while being faster and much cheaper than opus 4.5”
David Sacksat 1:15:15 · All-In1:15:15

Takeaways

  1. Coinbase fine-tuned an open model for fraud detection.
  2. The model cut fraudulent transactions by about 30% in tests.
  3. Specialist models can use less compute than frontier models.
Watch 2:07 from 1:14:54

In this episode

Full episode
  1. 5:58CoreWeave’s main risk is refinancing its fixed-price contracts7:10
  2. 31:11AI can lower software prices while expanding software demand6:32
  3. 53:40Anthropic should prioritize compute runway over IPO valuation6:13
  4. 40:19Payback period is the key data-center risk metric2:56
  5. 16:17LoudCloud’s bandwidth contracts show infrastructure risk2:14

  6. 1:14:54A small model cut Coinbase fraud by about 30%2:07

  7. 1:00:59OpenAI should have reset governance when it went for-profit2:04Up next

Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

  1. 1:14:58

    how Coinbase had fine-tuned an open-weight small model. This was QAN 3.5, the 9 billion parameter model, which is actually, for our audience, that's actually not a lot of parameters. It's not going to fit in your Roomba. Yeah, you could run it on your home computer. In any event, Coinbase fine-tuned it to find fraud. In testing, it cut fraudulent transactions by about 30% while being faster and much cheaper than opus 4.5 which is one of anthropic's better models and so what you're seeing in the enterprise right now is that for a lot of enterprise workloads you don't need a general super intelligence you don't need you know of course a frontier model you need a small specialist model that's been trained on your own data that you own and then you have a sovereignty you're not educating right frontier model right and these these small models are going to get more and more

  2. 1:15:55

    important as compute gets tighter and tighter, right? Because what you're doing is For a given GPU, you're essentially squeezing out the most performance, again, because when you run a huge model on kind of a mundane workload, the bigger the model, the more expensive it is to run, the more computer it needs. It's completely wasteful. Yes, exactly. So I think where the enterprise is headed right now, or at least it's a major, major trend, is that every workload is going to have its own small, customized model, right?

  3. 1:16:26

    That uses a fraction of the compute and energy. And, you know, and so you're starting to see this trend where now small models are taking share and lo and behold, coincidentally, Anthropic is now starting a scare campaign against small models. There you go. And again, and if you're concerned about the safety risks, the X risks that they're always talking about, small models are the most safe ones.

  4. 1:16:49

    The Terminator is not going to be run by a small model. It doesn't have that capability. So now they're trying to scare you about these completely safe, low powered,