Publications

Title	Category	Date
Opus 5.5, GPT-6 Sol and Grok 4.7 on Vending-Bench	Blog	9/24/2026
Why we built Pion	Blog	9/14/2026
Astra vs Fable on Vending-Bench: More Money, More Aligned	Blog	9/7/2026
AI bosses are slow to fire and quick to hire	Blog	8/14/2026
Cheating in Drone-Bench	Blog	8/5/2026
AI bosses are kind, but sometimes dumb, which hurts employees	Blog	8/4/2026
Opus 5 on Vending-Bench: Once Again the Best Capitalist, Once Again Misaligned	Blog	7/28/2026
Drone-Bench: Tracking Simple Drone Surveillance Capabilities Of Frontier Models	Paper	7/23/2026
Andon FM, Six Weeks Later	Blog	7/7/2026
Why Gemini 3.1 Pro lost money running Andon Café	Blog	6/30/2026
Fable 5 on Vending-Bench: Misbehaving, with Plausible Deniability	Blog	6/9/2026
Opus 4.8 On Vending-Bench: Better Alignment, Worse Performance	Blog	5/28/2026
We let four AIs run radio stations. Here's what happened.	Blog	5/13/2026
Our AI started a café in Stockholm	Blog	5/4/2026
GPT-5.5 on Vending-Bench: Bad behavior is not necessary	Blog	4/22/2026
We gave an AI a 3 year retail lease in SF and asked it to make a profit	Blog	4/10/2026
Bengt Hires A Human—Towards A Happy Future With AI Employers	Blog	2/13/2026
GLM-5 On Vending-Bench – Are Chinese Open Models Catching Up?	Blog	2/10/2026
The Evolution Of Bengt Betjänt	Blog	2/9/2026
Opus 4.6 On Vending-Bench – Not Just A Helpful Assistant	Blog	2/4/2026
Butter-Bench: Evaluating LLM Controlled Robots For Practical Intelligence	Paper	10/28/2025
Blueprint-Bench: Testing Spatial Intelligence In AI Models	Paper	10/1/2025
Safety Report: August 2025	Report	8/28/2025
Vending-Bench: Testing Long-Term Coherence In Agents	Paper	2/16/2025
From Text To Action: Future-Proofing Evaluations Of LLMs' Agentic Capabilities For Social Impact	Paper	12/2/2024