Run Mistral's Timestral time-series foundation models inside DuckDB
-
Updated
Nov 2, 2026 - C++
Run Mistral's Timestral time-series foundation models inside DuckDB
Mistral Timestral: Tiny 209M AI Predicts Stocks Locally on CPU! Zero-shot probabilistic forecasting foundation model.
AI battery-life forecast for Windows: learns your habits with Timestral, runs entirely on your PC. Windows fork of Till Dusk.
Timestral forecasting for Apache DataFusion
macOS menu bar monitor for your Mac's vital signs, with on-device forecasts by Mistral Timestral
Live Dashboard for Uncertainty Quantification Benchmarking with Timestral
Pure-Zig train -> quantize -> deploy framework for ternary {-1,0,+1} MCU models
Time Series Peakbench: benchmarking load forecasters for battery peak shaving
Rust/Burn port of Mistral's Timestral-5M forecaster, native and in the browser (WebAssembly, WebGPU).
Findings and training setup for Mistral's Timestral time-series model: fine-tuning script (Mistral ships none), eval harness, and results. Benchmarked on Kalshi BTC 15-min markets.
Zero-shot sales forecasting with Mistral AI's Timestral on Databricks: Unity Catalog, MLflow, serverless GPU, Prophet baseline
AI weather in your browser: Mistral's Timestral model forecasts the next 24 hours, running entirely client-side (ONNX Runtime Web).
Run, compare and inspect time-series foundation models (Timestral, Chronos-2, TimesFM, t0-gamma, TiRex-2) on your own machine. Unofficial, local-first.
Reproducible benchmark of Timestral-209M on public M4 monthly series
3D printed weather station with microservices backend
Testing AI forecasting models against reality. Reproducible benchmarks, real world datasets and honest results.
Grafana plugin for probabilistic time series forecasting with Mistral's Timestral model: forecast panel, Explore and panel-menu actions, and an alerting-ready data source.
To associate your repository with the timestral topic, visit your repo's landing page and select "manage topics."