Sold by: Tractian
Deployed on AWS
JANUS is TRACTIAN's industrial reasoning model for machine diagnostics and reliability engineering. Delivered via an OpenAI-compatible chat completions API, JANUS accepts raw vibration, current, and process signals and returns structured verdicts with fault identification, root cause, severity levels, time-to-failure estimates, confidence scores, and evidence-backed maintenance recommendations. Trained on TRACTIAN's proprietary dataset of millions of real machine signals, JANUS outperforms frontier LLMs by 47% and certified human analysts by 93% on industrial diagnostic accuracy benchmarks.
Overview
JANUS is a vision-language model and agent harness purpose-built for industrial machines by TRACTIAN, named to the Forbes AI 50 list.
The system combines two components: (1) a vision-language model with TRACTIAN-owned weights, trained exclusively on industrial machine data and running on private US infrastructure, and (2) an agent harness where a lead coordinator dispatches domain-specific expert agents for mechanical, electrical, and operational analysis.
Core Capabilities:
- Reliability: Fault diagnosis and component-level root cause analysis from raw sensor signals
- Maintenance: Prioritized work recommendations with urgency scoring
- Time-Series Analysis: Anomaly detection, trend analysis, and change-point detection
- Process Analysis: Identification of inefficiencies and deviations across multi-stage operations
- Prediction: Time-to-failure and degradation forecasting
- Recommendation: Next-best-action with complete evidence chains
Integration: JANUS exposes an OpenAI-compatible endpoint (POST /v1/chat/completions, model: janus-1) supporting structured JSON response formatting. Integration requires minimal code changes for teams already using OpenAI or Anthropic client libraries.
Data Handling: All inference runs on private US-based infrastructure. No customer data is retained after processing. No third-party models are used in the inference pipeline.
Benchmark Performance (5 diagnostic cases):
- 89.8% average diagnostic accuracy
- +47% accuracy vs. frontier LLMs (GPT-4, Claude)
- +210% consistency vs. frontier LLMs
- +93% accuracy vs. certified human vibration analysts
- +450% consistency vs. certified human analysts
Highlights
- Industrial Reasoning Model: 89.8% diagnostic accuracy, outperforming frontier LLMs by 47% and certified human analysts by 93%. Purpose-built on TRACTIAN's proprietary dataset of millions of real machine signals.
- OpenAI-Compatible API: Drop-in integration via standard chat completions endpoint (model: janus-1). Returns structured JSON with fault diagnosis, severity, time-to-failure, confidence scores, and evidence-backed maintenance actions.
- Private US Infrastructure: All inference on TRACTIAN-owned US compute. Zero data retention after processing. No third-party models in the pipeline. Full data sovereignty for regulated industries.
Details
New
Introducing multi-product solutions
You can now purchase comprehensive solutions tailored to use cases and industries.
Features and programs
Financing for AWS Marketplace purchases
AWS Marketplace now accepts line of credit payments through the PNC Vendor Finance program. This program is available to select AWS customers in the US, excluding NV, NC, ND, TN, & VT.
Pricing
Pricing is based on actual usage, with charges varying according to how much you consume. Subscriptions have no end date and may be canceled any time.
Additional AWS infrastructure costs may apply. Use the AWS Pricing Calculator to estimate your infrastructure costs.
Dimension | Description | Cost/unit |
|---|---|---|
Input Tokens (1M) | Input tokens processed, billed per million tokens | $12.50 |
Output Tokens (1M) | Output tokens generated, billed per million tokens | $35.00 |
Dimensions summary
You pay based on usage, split into two separate dimensions. Input Tokens cover the data JANUS processes, billed per million tokens. Output Tokens cover the responses JANUS generates, also billed per million tokens. Tokens are the units of text the model reads and produces. Your total cost scales with how much you send in and how much the model returns. These two dimensions are independent, so heavier input or heavier output each adds to your bill separately. There is no fixed commitment or seat count; you pay for what you use.
Top-of-mind questions for buyers
A token is a small piece of text the model reads or writes, roughly a word fragment. Input tokens include your request and the asset context JANUS pulls in, like signals and world model data. Output tokens include the structured verdict, evidence chain, and recommendation the model returns. Both are counted per million.
Both charges apply independently and appear together. Input tokens tend to dominate because each request grounds the model in asset signals, sensor streams, and world model context server-side. Output stays compact since responses return a structured verdict with an evidence chain. Diagnosing many assets with rich context raises input volume most.
One endpoint covers six tasks: reliability, maintenance, time-series analysis, process analysis, forecasting, and recommendation. A single request can trigger the agent pipeline that reviews evidence across domains. More reasoning and more asset context both increase input tokens. The returned verdict and evidence chain add to output tokens.
tractian.com
Helpful?
Vendor refund policy
Usage charges for TRACTIAN JANUS are billed through AWS Marketplace and are non-refundable once incurred, except as follows. If you were billed in error, or an incident on our side prevented you from using the API, email support@tractian.com within 30 days with your AWS account ID and the charge in question. We reply within one business day and, for valid claims, ask AWS Marketplace to refund the account charged. Cancel anytime on your AWS Marketplace subscriptions page.
Legal
Vendor terms and conditions
Upon subscribing to this product, you must acknowledge and agree to the terms and conditions outlined in the vendor's End User License Agreement (EULA) .
Content disclaimer
Vendors are responsible for their product descriptions and other product content. AWS does not warrant that vendors' product descriptions or other product content are accurate, complete, reliable, current, or error-free.
Delivery details
Software as a Service (SaaS)
SaaS delivers cloud-based software applications directly to customers over the internet. You can access these applications through a subscription model. You will pay recurring monthly usage fees through your AWS bill, while AWS handles deployment and infrastructure management, ensuring scalability, reliability, and seamless integration with other AWS services.
Resources
Vendor resources
Support
Vendor support
Email: support@tractian.com
Scope: Onboarding, API integration, usage, troubleshooting, and billing inquiries related to the TRACTIAN JANUS listing on AWS Marketplace.
Response time: First response within one business day.
Languages: English, Portuguese, Spanish.
AWS infrastructure support
AWS Support is a one-on-one, fast-response support channel that is staffed 24x7x365 with experienced and technical support engineers. The service helps customers of all sizes and technical abilities to successfully utilize the products and features provided by Amazon Web Services.
Similar products
Vendor-agnostic predictive maintenance service combining IoT sensors from Augury, Tractian, I-care, Fluke, and AssetWatch with AWS cloud ML analytics to detect equipment anomalies and reduce unplanned downtime.
Predictive maintenance and reliability engineering are essential for smooth manufacturing operations. Our solution streamlines every phase of the manufacturing lifecycle—from design to sustainability—using cutting-edge technologies like IoT, AI, and digital twins. This suite enhances efficiency, reduces waste, and drives smarter manufacturing decisions. Leveraging Industrial Data Fabric on AWS, operations leaders can predict equipment failures, optimize maintenance schedules, and enhance asset reliability. By integrating data from various sources, this solution empowers predictive maintenance and reliability initiatives across all stages, ensuring sustainable and resilient production processes. ATTENTION MONITRON CUSTOMERS: Amazon Monitron will no longer be available to new customers effective October 31, 2024. If you're impacted, please scroll down to the support section or reach out to leverage our Amazon Monitron Transition Plan, offered at no cost to you.
Customer reviews
No customer reviews yet
Be the first to review this product . We've partnered with PeerSpot to gather customer feedback. You can share your experience by writing or recording a review, or scheduling a call with a PeerSpot analyst.