Free Systems Lab Turns 476,000 Words of Model Cards Into 28 Comparable Frontier Cards
soumitrashukla9 · x · 2026-09-11
- Free Systems Lab built 28 frontier AI model cards from scratch, working through 476,000 words of model and system-card material, and shipped a Model Cards explorer to make the frontier's public record easier to browse and compare.
- Key finding: published evidence across labs rarely lines up cleanly, with gaps in safety and capability reporting.
- Cards cover OpenAI, Anthropic, Google DeepMind, Meta, Mistral, xAI, Thinking Machines, Moonshot AI, DeepSeek and more. A companion post by ahallresearch describes the goal: condensing cards to "no larger than a Pokémon card," early-stage work with feedback welcome.
More from Models
- DeepSeek v4.1 Flash Spotted Online, Authenticity Unverified — petrusenko_max · 2026-09-11
- Persimmon team members share months-in-the-making launch, research preview open — niloofar_mire · 2026-09-11
- Power user: Astra's usage limits are 'a joke' compared to Google's plan — MickeySteamboat · 2026-09-11
- GPT Astra Takes on Dominions 6, a Brutally Complex 4X Strategy Game — garden_frog · 2026-09-11
- Benchwarmer Tool Rebuilds Misleading AI Benchmark Charts and Recomputes the Winners — aronchick · 2026-09-11
- GPT-6 Astra autonomously flies a drone to find and follow a person, tops Drone-Bench — TheMoonMidas · 2026-09-11