githug · Blog
What is the best model for building time-series forecasting applications — google/timesfm-3.0-pytorch or another solution?
Downloads24M/mo
Accessopen · no token
Tasktime-series-forecasting
Updated3mo ago
Downloads8.9M/mo
Accessopen · no token
Tasktime-series-forecasting
Updated3mo ago
Downloads444k/mo
Accessopen · no token
Tasktime-series-forecasting
Updated5d ago

When it comes to time-series forecasting, choosing the right model is akin to selecting the right compass for a voyage: the right choice can steer you toward reliable insights while the wrong one can leave you lost. In the battle of models, Amazon's Chronos-2 emerges as a leading contender over Google's TimesFM-3.0, backed by a larger adoption and solid maintenance.

Top Model Comparison

Model Downloads Likes License Last Updated Link
amazon/chronos-2 23,768,593 435 Apache-2.0 Jun 5, 2026 ✅ healthy
autogluon/chronos-2 8,879,175 48 Apache-2.0 Jun 5, 2026 ✅ healthy
google/timesfm-3.0-pytorch 444,052 600 Other Sep 2, 2026 ⚠️ caution

Verdicts:

Recommendation:

Choose Amazon Chronos-2 if you need a well-adopted, actively maintained solution with a clear license. Opt for Google TimesFM-3.0 only if you have strong reasons to use its specific features, but be wary of potential licensing issues.

🤗 huggingface.co/amazon/chronos-2🤗 huggingface.co/autogluon/chronos-2🤗 huggingface.co/amazon/chronos-bolt-small🤗 huggingface.co/autogluon/chronos-bolt-small🤗 huggingface.co/autogluon/chronos-2-small🤗 huggingface.co/google/timesfm-3.0-pytorch
Answered live from GitHub & Hugging Face · openai/gpt-4o-mini · 9/8/2026
Ask your own question →