Abel AI Lab open-sources Simons time series foundation models, up to 297M params
NielsRogge · x · 2026-11-18
Abel AI Lab released Simons, a family of open time series foundation models built with automated research. Weights, training code, and public training data are all open.
- Four sizes: 25M / 60M / 178M / 297M params, with context up to 32,768 time steps
- Zero-shot probabilistic forecasting out of the box
- The 25M model ranks #3 on the TIME leaderboard on Papers with Code, ahead of Chronos-2 and TimesFM 2.5
- Paper, GitHub code, Hugging Face models/datasets, and an online Forecast Explorer are all available
Notably, it's a public example of models produced by automated research pipelines.
More from Models
- Gemini side panel in Chrome leaks model tiers: free vs Pro subscription compared — gaganghotra_ · 2026-11-18
- Practitioner: no AI bubble, frontier labs lead open source 10x on knowledge work — gabriel1 · 2026-11-18
- Gemini cracks an unread 1466 Sforza cipher by hunting down a 1465 key in Vienna — fofrAI · 2026-11-18
- Cloudflare's Clairic extraction model returns confidence scores and spans, no preset categories — altryne · 2026-11-18
- Claude's biggest mistakes come from following rules too rigidly — wavefnx · 2026-11-18
- Interact-0: a 31B full-duplex speech model that knows when to speak in groups — niloofar_mire · 2026-11-18