The language model of the AETRON network
A large language model that runs on top of AETRON. Every answer can be re-checked cryptographically. The Pre-Alpha is already live on mainnet.
Model parameters
Core capabilities
Multimodal understanding
Working with text, images and documents in one context
- Understanding of long text
- Image analysis and recognition
- Pulling data out of documents
- Tying different input types into one request
Working with code
Writing, reviewing and debugging code in different languages
- Support for the popular programming languages
- Optimisation suggestions
- Spotting and fixing errors
- Hints about project architecture
Research
Help with sources, data and academic writing
- Searching and matching sources
- Working with data and visualisations
- Pointers on methodology
- Structuring an academic paper
Text and content
Drafting and editing text in different genres
- Preparing and reworking content
- Fiction and scriptwriting
- Marketing copy
- Keeping a single tone and style
Benchmarks
What the AETRON network adds
Distributed inference
The model runs across thousands of GPUs in the participant network - not tied to a single data center. This removes the capital barrier and eliminates the single point of failure.
Verifiable privacy
Sensitive workloads can use the optional Confidential Computing mode with TEE attestation. The customer gets a guarantee that the data didn't leak, without trusting the operator.
A path to self-improving AI
AETRON is architecturally ready for a scenario where AI agents run their own research and train new models. Every step of that process is recorded on-chain and can be re-checked.
Free access to inference
At the protocol level no one pays for requests. Miners earn through AET emission, the end user gets the model for free.
Where it fits
Development
- Code review and optimisation
- API documentation
- Drafting test scenarios
- Architecture sketches
Business analytics
- Working with reports and data
- A round-up of relevant research
- Help with making decisions
- Risk assessment
Content and communications
- Technical documentation
- Marketing copy
- Emails and newsletters
- Blog material and SEO
Integration via API
REST API
Plain HTTP endpoints for integration with existing applications
COMING SOONWebSocket streaming
Streaming responses for chats and interactive interfaces
COMING SOONSDK
Client libraries for the popular languages, with error handling and retries built in
COMING SOON
Get early access
AETRON LLM is in Pre-Alpha and live on mainnet. Sign up to get priority access and updates on what comes next.