Cisco Time Series Model 1.0 Preview
Model directorycisco-ai/cisco-time-series-model-1.0-preview500M params | Upstream checkpoint
The Cisco Time Series Model 1.0 Preview is Cisco's 500M-parameter univariate zero-shot forecasting model, tuned for observability-style workloads where coarse and fine temporal structure both matter. As the preview release in the family it pairs the largest parameter count with a deep stack of 50 transformer layers.
Architecturally it is a multiresolution decoder-only transformer initialized from TimesFM weights, though the official card explicitly notes that it does not conform to the TimesFM architecture; the multiresolution design is what lets it capture both fine-grained and coarse-grained patterns, and it emits 9 quantile outputs for probabilistic forecasts. Pretraining draws on 300B+ points total, with the card attributing more than 50% to Splunk Observability Cloud metrics, 30% to GiftEvalPretrain, 5% to Chronos datasets, and 15% to synthetic multiresolution data.
Where to run this model
Compute availability, hardware requirements, and pricing depend on the execution provider.
Get model weights, deployment instructions, and current provider listings.
Hugging Face provider mappings checked 2026-10-06. No Inference Provider mapping was listed for this checkpoint.
Model Classification
Family
Cisco Time Series Model
Type
time series foundation model
Pretrained time-series model for zero-shot or few-shot forecasting workloads.
Resources
Training Data
300B+ points total, with the official card attributing more than 50% to Splunk Observability Cloud metrics, 30% to GiftEvalPretrain, 5% to Chronos datasets, and 15% to synthetic multiresolution data.
Recommended For
- • Infrastructure monitoring and network telemetry forecasting
- • Observability workloads where multiresolution context improves accuracy
Strengths
- • Built on 300B+ observability data points from Cisco/Splunk production infrastructure
- • Multiresolution architecture captures both fine-grained and coarse-grained patterns
Limitations
- • Domain-specific training may limit generalization to non-infrastructure series
- • Preview release — behavior and API surface may evolve
Capabilities
Tags
Specifications
- Parameters
- 500M
- Architecture
- decoder-only transformer (multiresolution; initialized from TimesFM weights)
- Context length
- n/a
- Max context
- n/a
- Minimum history
- n/a
- Recommended history
- n/a
- Input step
- n/a
- Required target series
- 1
- Temperature
- Ignored
- Top P
- Ignored
- Max output
- n/a
- License
- n/a