# Foundation Models Unlock Time Series Scaling
_Toto 2.0 foundation models demonstrate remarkable scaling, achieving state-of-the-art forecasting performance across multiple benchmarks with a unified training approach._
**Published:** 2026-05-20
**Source:** https://www.startuphub.ai/ai-news/ai-research/2026/foundation-models-unlock-time-series-scaling
---
The promise of foundation models has largely been confined to NLP and vision, leaving the critical domain of time series forecasting in a fragmented state. This work demonstrates that time series models, much like their counterparts in other domains, exhibit remarkable scalability, with a single training recipe yielding consistent forecast quality gains from millions to billions of parameters. The researchers behind [Toto 2.0](https://arxiv.org/abs/2605.20119v1) have codified this insight into a practical framework.

Time Series FragmentationDriver

From the article 4 mentionsThe promise of foundation models has largely been confined to NLP and vision, leaving the critical domain of time series forecasting in a fragmented state.addressed byToto 2.0 Foundation ModelsCorenew foundation models demonstrate remarkable scalability for time seriesFrom the article 3 mentionsThe five Toto 2.0 forecasting models released under Apache 2.0 are a testament to this unified approach, setting new benchmarks in forecast quality.Unified Scaling RecipeCoresingle training approach effective across millions to billions of parametersApache 2.0 ReleaseEffectFrom the articleThe five Toto 2.0 forecasting models released under Apache 2.0 are a testament to this unified approach, setting new benchmarks in forecast quality.Consistent Quality GainsEffectforecast quality improves reliably with increased model parameter sizeFrom the articleThis work demonstrates that time series models, much like their counterparts in other domains, exhibit remarkable scalability, with a single training recipe yielding consistent forecast quality gains from millions to billions of parameters.Practical FrameworkEffectcodified insights into a usable and accessible framework for researchersFrom the articleThe researchers behind Toto 2.0 have codified this insight into a practical framework.leads toState-of-the-Art PerformanceOutcomeachieving new benchmarks across multiple forecasting benchmarksFrom the articleThe Toto 2.0 forecasting models have established new state-of-the-art results on three distinct forecasting benchmarks: BOOM (observability), GIFT-Eval (general-purpose), and the contamination-resistant TIME benchmark.

## Unified Scaling Recipe for Forecast Accuracy

The core innovation lies in a robust training methodology that proves effective across a wide spectrum of model sizes, from 4 million to 2.5 billion parameters. This scaling law suggests a path towards highly performant and reliable time series forecasting without the need for bespoke tuning for each parameter class. The five [Toto 2.0 forecasting models](https://arxiv.org/abs/2605.20119v1) released under Apache 2.0 are a testament to this unified approach, setting new benchmarks in forecast quality.

## State-of-the-Art Performance Across Benchmarks

The [Toto 2.0 forecasting models](https://arxiv.org/abs/2605.20119v1) have established new state-of-the-art results on three distinct forecasting benchmarks: BOOM (observability), GIFT-Eval (general-purpose), and the contamination-resistant TIME benchmark. This broad success underscores the generalizability of the architecture and training recipe, addressing a key challenge in the time series domain. The report details not only the experimental outcomes but also the architectural design, training data strategy, and the innovative u-muP hyperparameter transfer pipeline that underpins these achievements.
---
Original analysis from [startuphub.ai](https://www.startuphub.ai), the #1 AI startup directory.