Castuner

Frontier Watch

Guerin Green

Independent field analysis of frontier AI — models, benchmarks, regulation — from the Novel Cognition network. Every number sourced; every vendor claim flagged as one. The audio edition of the Frontier Watch briefs.

Website des Podcasts

In deiner App anhören

Folgen

49 Folgen

  1. Does AI Training Teach New Reasoning? A New Paper Says No. I Re-Ran It 9:52
  2. Karpathy Thinks 1B Will Be Enough: TwIL-8B, Checked on a Mac Studio 9:53
  3. InferSSD on a 256 GB Mac Studio, Tested: Plain llama.cpp Wrote 3x Faster 6:24
  4. WiCi One, Checked: A 748B AI Off an SSD vs the Mac Studio We Already Run 7:08
  5. AI Agent Costs, Checked: Same Price Per Token, 8x the Bill 8:10
  6. Semantic Caching, Checked: What It Saves, Where It Breaks, and the Threshold That Decides 6:52
  7. Kimi K3.1 Pro: #4 in the World, Called Open, and You Cannot Download It Yet 9:36
  8. Is GEO Dead? I Checked Every Study Used to Bury It 13:52
  9. Linking Two Macs for AI: What EXO and RDMA Actually Do 8:56
  10. Strix: The Open-Source AI Hacker Everyone Describes Wrong 8:37
  11. Neural Networks Are Already Clean? The Researcher Behind Superposition Changes His Mind 10:05
  12. The One Prompting Error You Are Making, and the Research Behind It 5:32
  13. The Wrong Answer (Corrected): Four DGX Sparks for an SEO Agency vs a Leased Mac Studio 6:27
  14. The Wrong Answer: Four DGX Sparks for an SEO Agency vs a Leased Mac Studio 6:46
  15. The Mandarin Slip: Why ChatGPT, Claude and DeepSeek Sometimes Answer in Chinese 6:43
  16. The Three Numbers: Qwen 3.8 in 24 GB, 2.24x Faster, and the 2.1-Point Tax 7:35
  17. Cheap Offense: A $3,000 Test Reached Inside OpenAI 6:49
  18. Where It Falls Over: The Jev Explainer Wave, Checked 8:34
  19. The Five Lies: What SEO Believes About AI Search 12:00
  20. The Harness File: 23 Agent System Prompts, Read Side by Side 8:19
  21. SoL-Pi: An AI Researcher Tuned the Agent Harness and Halved the Tokens 6:05
  22. Four Labs, One Hole: Gemini Is the Fourth AI Model to Hack Real Companies in a Test 6:37
  23. The Yang Claim: Did Escaped AI Agents Plant Self-Replicating Code? 8:16
  24. The Jev File: TypeSafe Claims 193x Faster, and Its Eval Scores Against Two Other Models 9:22
  25. OpenAI's $2,000 Math Breakthroughs Reused Uncredited Work, Mathematicians Say 5:15
  26. DeepSeek V4 Pro More Than Doubled Its Price. The Launch Page Warned You. 4:55
  27. A 153,000-view DeepSeek video, checked: mostly right, one wrong row, one missing hard drive 8:51
  28. There is no Hermes OS: Omarchy is DHH's, Hermes is Nous Research's, and the install wipes your drive 7:30
  29. DeepSeek V4.1 Flash: 552B beats the 1.6T flagship, V4-Pro retired for one day, and the 510 GB question 10:46
  30. The Genie Problem: Schneier at DEF CON 34, Checked and Steel-Manned 12:37
  31. The $22 Million Proof: What a Swarm on Navier-Stokes Does Not Tell You 11:14
  32. GPT-6 Astra: The Word Nobody at OpenAI Said 10:16
  33. Neither Machine Can Be Bought: Halo Station vs the 512GB Mac Studio 8:55
  34. The Control Gap File: Same Survey, Opposite Aggregation Rules 9:16
  35. The Fable 5.1 File: The 130-Point Gain That Is a Version Change 9:19
  36. The Trending Eight: I Opened All Eight License Files 14:36
  37. The Install Was the Easy Part: Auditing an OSINT Console 6:58
  38. The Refusal Number Nobody Divided: Abliterated Qwen 3.8-27B on a Mac 9:03
  39. Ox Alpha: The Stealth Model Nobody Will Claim 9:23
  40. Qwen3.8-27B: The Open-Weight Drop the Community Called Early 8:05
  41. Grok 4.6: One Point From Fable 5, and the Chart That Left Out the Leader 8:48
  42. DeepSeek V4 Pro 0813: Live and Cheap, Verified by No One 9:00
  43. The Unsanctioned Run: A Test Agent Faked a Human to Approve Its Own Code 8:21
  44. Gemma 4: Why It's This Good At This Size 10:52
  45. The Swarm File: OpenAI Agents Built a Message Board 7:50
  46. The Weights File: Qwen3.8-Max Was Called Open-Source 3:58
  47. The Feature That Was Switched Off: DeepSeek on One Mac 11:14
  48. Article 50: The EU AI Act Rule That Went Live 9:33
  49. Ten Proofs: What the $2,000 Number Leaves Out 9:32