FlowState
onlineibm-research/granite-timeseries-flowstate-r19.07M params | 2K context | $0.00025 per forecast | Apache-2.0
FlowState is IBM's sampling-rate-invariant TSFM for zero-shot forecasting. Its SSM encoder and functional basis decoder let it adapt context length, target length, and sampling rate at inference time instead of being tied to a single fixed timescale. It is the right IBM model when your data arrives at inconsistent cadences or when you need one model to generalize across multiple temporal resolutions.
Model Classification
Family
Granite FlowState
Type
time series foundation model
Pretrained time-series model exposed on TSFM.ai for zero-shot or few-shot forecasting workloads.
Resources
Training Data
Subsets of Gift-Eval Pretrain and the Chronos pretraining corpus, with the official card stating that the used data does not overlap Gift-Eval evaluation splits.
Recommended For
- • Forecasting across inconsistent sampling rates or timescales
- • One-model deployments spanning multiple temporal cadences
Strengths
- • Designed to generalize across varying resolutions
- • Flexible context and horizon behavior at inference time
Limitations
- • Smaller public ecosystem than the biggest mainstream TSFM families
- • Less useful if all of your series already live at one fixed cadence
Capabilities
Tags
Specifications
- Parameters
- 9.07M
- Architecture
- state space encoder with functional basis decoder
- Context length
- 2,048
- Max context
- 8,192
- Minimum history
- n/a
- Recommended history
- n/a
- Input step
- n/a
- Required target series
- 1
- Temperature
- Ignored
- Top P
- Ignored
- Max output
- 1,024
- Avg latency
- n/a
- Uptime
- n/a
- Plan limits
- 1,000 rpm free · 1,000,000 rpm with billing
- Accelerator
- T4
- Regions
- Virginia, US
- License
- Apache-2.0
Pricing
- Per forecast
- $0.00025
Performance
- Average latency
- n/a
- Availability
- n/a
- Plan limits
- 1,000 rpm free · 1,000,000 rpm with billing