Sage-A1.5 is a compact foundation model that Stratat has made available for local, offline inference on desktops through its Strata engine. It is meant for users who want private, on-device intelligence, so you can run it inside the Stratat client without needing a datacenter. The model was released under the InternScience organization as Sage-A1.5, and Stratat provides a static-page delivery that includes usage instructions and measurement guidance.
What Sage-A1.5 is and why it exists
The Sage-A1.5 model is a small language model intended for edge use. Its purpose is to let people work with text intelligence on their own machines instead of sending queries to remote services. Because it runs locally, it keeps your prompts and outputs on your computer rather than in the cloud.
Hardware requirements to run Sage-A1.5 locally
The model was built for desktop GPUs and CPUs that meet Stratat's supported specs. Typical supported setups include modern consumer GPUs (for example Nvidia cards with CUDA support) and multi‑core x86 CPUs together with at least 8 GB of RAM. The Strata engine compiles the model for the platform, so the same file can be loaded on any machine that satisfies those baseline conditions.
How to verify the model runs on your own machine
The easiest way to check that Sage-A1.5 works on your system is to load the model card from the Stratat site and click the provided local‑run button. After launch, send a short prompt such as "translate this sentence" to the model and confirm that a response appears without any network activity. You can also open the Strata engine logs panel and look for a line that shows the model size being loaded and the device ID reported for inference.
| Component | Measured value |
|---|---|
| Peak memory use during inference | ~6 GB |
| Average generation latency (short prompt) | ~3.5 s |
| Recommended GPU compute capability | 8.6 |
Sage-A1.5 is defined as a small language model that fits on consumer‑grade hardware and can be executed entirely offline, using the Strata engine for fast token generation.
Where to find the model and run tests
The model page on the Stratat site lists Sage-A1.5 alongside other released models and includes a download link plus a quick start guide. Follow the guide to install the desktop client, launch the app, and press the run button to test loading the model on your own machine.
Why pick Sage-A1.5 over other small models
Sage-A1.5 balances size and quality better than many other tiny models released recently. It can handle normal conversation, summarisation, and simple reasoning tasks while staying small enough to fit on a single consumer GPU.