Google releases TimesFM 3.0: 330M parameter zero-shot multivariate forecasting model
osanseviero · x · 2026-09-01
Google Research has released TimesFM 3.0, an open-source foundation model for time series forecasting. With 330 million parameters, it introduces native support for multivariate forecasting in a zero-shot manner, overcoming the univariate limitation of previous versions. Pre-trained on over 1 trillion time points, TimesFM-3 significantly outperforms other models across major benchmarks.
Related event: Google Releases TimesFM 3.0 Open-Source Multivariate Forecasting Model(3 posts)→
More from Models
- Bitsec subnet outperforms Anthropic's hardened Fable 5 in bug detection — markjeffrey · 2026-09-01
- Qwen 3.8 27B with RLMs performs reliably for local deployment — generativist · 2026-09-01
- Observation: Qwen Agents Make Fewer Topology Errors in Chinese Contexts — cephaloform · 2026-09-01
- Claude Opus 5 accused of over-executing prompts, doing unnecessary work — Own-Adhesiveness-705 · 2026-09-01
- AI writes perfect English, yet nobody enjoys reading it — imperfection is the point — auto_grad_ · 2026-09-01
- Fable 5.1 spotted staged on Bedrock, launch may be days away — haider1 · 2026-09-01