GGUF models
Manage models in the GGUF format within the same operational catalogue.
Model Catalogue
Bring the right model to every AI workload.
The T-Flux Ultra Model Catalogue provides a central catalogue and operational management view of every AI model available to T-Flux.
It manages AI intelligence resources separately from platform infrastructure and tenant operations, so teams can see what models are available, how they are performing, and how T-Flux is using them.
Open standardsModels chosen for your environment.
Formats, families, private endpoints, and customer foundation models.
GGUFSafeTensorPrivateBYO
What models are available? How are they performing? How is T-flux Ultra using them?
Flexible by design
Use the models that best serve the customer environment.
The Model Catalogue gives AI administrators, ModelOps teams, platform engineers, and authorised technical administrators a single operational view of compute and models, with fast drill-down to request, job, and document evidence.
Catalogue management
Manage and control the available models and their live operational characteristics.
Registered catalogue
Model type and architecture, format, size and parameter count, version and lifecycle status, deployment status, and hosting location.
Live characteristics
GPU and memory requirements, active instances, inference throughput, tokens per second, context-window utilisation, latency, and error rate.
Assignments and routing
Tenant and model assignments, routing rules, model utilisation, routing decisions, and the role each model plays in an AI request.
Governance and lifecycle
Model provenance and approval, versioning, retirement, replacement, and controlled visibility of governance activity.
Multi-model orchestration
Make every routing and synthesis decision visible.
For multi-model orchestration, the Model Catalogue shows how models contribute to a request, how their responses compare, and how T-Flux selects or synthesises the final result.
Participation
See which models participated in a request and which model produced each candidate response.
Selection
Understand model agreement or disagreement, the synthesis process, arbitration time, and which response was selected.
Performance
Compare model-specific throughput, tokens per second, GPU seconds per request, latency, error rate, citation coverage, and refusal rate.
Customer benefits
Get more from T-Flux Ultra with a catalogue built for model choice and control.
Choice without lock-in
Open formats, supported model families, private endpoints, and customer foundation models give customers flexibility in how they build their AI capability.
Operational confidence
Live deployment, performance, resource, and failure measures help AI administrators and ModelOps teams understand model behaviour.
Governed model use
Provenance, approval, versioning, lifecycle status, and role-based visibility keep model decisions reviewable.
Better workload decisions
Comparison and routing views help teams understand which models deliver the strongest outcome for each request.
Role-based visibility
Give the right people the view they need.
Customer administrators can view model cards, export reports, and review governance logs. Operations teams can drill into operational cards. Sensitive fields, including API secrets, raw document snippets, and personally identifiable information, are masked by default.