timesfm

{"problem_it_solves": "Enables accurate time-series forecasting without requiring users to train models from scratch, supporting various forecasting horizons and contexts.", "target_customer": "Data scientists, ML engineers, researchers, and developers working on time-series forecasting tasks", "use_cases": ["Time-series forecasting", "Business forecasting", "Financial predictions", "Energy demand forecasting", "Any temporal prediction tasks"], "pricing_details": "Open source - Apache 2.0 license", "free_tier": "Yes - open source", "differentiator": "Google Research-backed foundation model with up to 16k context length and built-in quantile forecasting capabilities", "why_now": "Released September 2025 (v2.5), with ongoing development including Flax version for faster inference and expanded documentation", "traction": {"notable_metrics": "14.3k stars, 1.2k forks, 99 watchers, 301 commits", "press_mentions": ["Paper published at ICML 2024"]}}