Cisco Time Series Model 1.0

online

cisco-ai/cisco-time-series-model-1.0

250M params | 512 context | $0.5000 input | $1.50 output | Apache-2.0

Cisco's 250M-parameter univariate zero-shot forecasting model is the GA release of the Cisco Time Series Model family. It uses a multiresolution decoder-only architecture with 25 transformer layers and 15 quantile outputs, trained from scratch on observability data. Smaller and more efficient than the 500M preview, with denser probabilistic coverage.

Model Classification

Family

Cisco TSM

Type

time series foundation model

Pretrained time-series model exposed on TSFM.ai for zero-shot or few-shot forecasting workloads.

Resources

HuggingFace Paper

Training Data

Trained from scratch on observability-domain data; the official card attributes the core corpus to Splunk Observability Cloud metrics with additional public time-series sources.

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

Capabilities

forecastingquantile-forecastingzero-shotobservabilitylong-contextmulti-resolution

Specifications

Parameters: 250M
Architecture: decoder-only transformer (multiresolution; trained from scratch)
Context length: 512
Max context: 512
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

Input / 1M tokens: $0.5000
Output / 1M tokens: $1.50

Performance

Average latency: n/a
Availability: n/a
Plan limits: 1,000 rpm free · 1,000,000 rpm with billing