PRODUCT · MODEL HUB
Model hub
Every model your organization can use, in one governed catalog, behind one contract and one audit trail. Policy decides what each model may touch; routing decides which one gets the task.
PROVIDER
TIER
DEPLOYMENT
MAY TOUCH
STATUS
13 models · frontier and open weight, configured per deployment
Fable 5.1
Frontier reasoning for the hardest drafting and analysis tasks.
Astra 6
The agentic workhorse: deep reasoning on long documents and missions.
- Gemini
Meridian Pro
Long-context analysis across large document sets and media.
Fable o4
Reasoning specialist for math-heavy and multi-step analysis.
- Gemini
Meridian Ultra
The deepest tier for research-grade synthesis and review.
- AIEvaluating
Volt 3
In evaluation against reasoning and retrieval benchmarks.
- MISTRAL AIEvaluating
Corsair Large
European frontier model in evaluation for regulated clients.
Terra 4 Open
Open weights on your infrastructure, for data that never leaves.
- deepseek
Delver R2
Open reasoning model for in-boundary analytical work.
- Qwen
Qamar 235B
Strong multilingual open model, tuned for Arabic-heavy work.
Terra 4 Scout
Lighter open weights for fast in-boundary steps at scale.
- MISTRAL AI
Corsair Open 8x22
Sparse open model balancing quality and serving cost.
- Qwen
Qamar Coder 32B
Open coding model for in-boundary engineering workflows.
One contract, one audit trail
However many models run behind it
No lab lock-in
Models swap in and out behind one agreement. Your workflows never notice the change.
Newest models, day one
When a better model ships, it enters evaluation immediately and the hub absorbs it.
Spend you can see
Cost per task, team and model, with budgets enforced by spend controls.
The hub supplies. Routing decides.
Model routing sends each task to the right model in the hub, balanced for intelligence, accuracy, cost and privacy.
Explore model routing →Every model, on your terms.
PRODUCT · MODEL HUB
Model hub
Every model your organization can use, in one governed catalog, behind one contract and one audit trail. Policy decides what each model may touch; routing decides which one gets the task.
13 models · frontier and open weight, configured per deployment
Fable 5.1
Frontier reasoning for the hardest drafting and analysis tasks.
Astra 6
The agentic workhorse: deep reasoning on long documents and missions.
- Gemini
Meridian Pro
Long-context analysis across large document sets and media.
Fable o4
Reasoning specialist for math-heavy and multi-step analysis.
- Gemini
Meridian Ultra
The deepest tier for research-grade synthesis and review.
- AIEvaluating
Volt 3
In evaluation against reasoning and retrieval benchmarks.
- MISTRAL AIEvaluating
Corsair Large
European frontier model in evaluation for regulated clients.
Terra 4 Open
Open weights on your infrastructure, for data that never leaves.
- deepseek
Delver R2
Open reasoning model for in-boundary analytical work.
- Qwen
Qamar 235B
Strong multilingual open model, tuned for Arabic-heavy work.
Terra 4 Scout
Lighter open weights for fast in-boundary steps at scale.
- MISTRAL AI
Corsair Open 8x22
Sparse open model balancing quality and serving cost.
- Qwen
Qamar Coder 32B
Open coding model for in-boundary engineering workflows.
One contract, one audit trail. Models swap in and out behind one agreement; the newest enter evaluation the day they ship, and spend is visible per task, team and model.