SignalTech/AI
H2O.ai's TimeH2O Beats Mistral, Google, and Amazon in Time-Series Forecasting
H2O.ai's TimeH2O outperforms Mistral, Google, and Amazon in time-series forecasting.
Context
What happened: H2O.ai released TimeH2O-v1, a time-series foundation model that ranks first of 32 models on the fev-bench leaderboard by win rate.
Key numbers: 91.9% win rate and skill score of 50.8 across 100 forecasting tasks; 108 million parameters; Timestral-209M follows at 89.6% and t0-gamma at 88.1%; leaderboard as of October 23, 2026.
Why it matters: H2O.ai says the zero-shot model, never trained on any fev-bench dataset, can be fine-tuned into a domain-specific foundation model for each customer's use case.
Available on Hugging Face under a non-commercial license.
Market: +0.2% today ($77.25), -4.6% YTD, 0.6x vol — as of 08:00 UTC
What changed
H2O.ai released TimeH2O-v1, a time-series foundation model that ranks first of 32 models on the fev-bench leaderboard by win rate.
Why it matters
H2O.ai says the zero-shot model, never trained on any fev-bench dataset, can be fine-tuned into a domain-specific foundation model for each customer's use case.
Available on Hugging Face under a non-commercial license.
Market: +0.2% today ($77.25), -4.6% YTD, 0.6x vol — as of 08:00 UTC