Instructions to use metafloor-ai/manifest-orchestrator-30b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use metafloor-ai/manifest-orchestrator-30b with PEFT:
Task type is invalid.
- Notebooks
- Google Colab
- Kaggle
Manifest 30B β coming soon
Part of the Manifest suite by MetaFloor β frontier, enterprise-scale supply-chain intelligence.
π Coming soon. Manifest 30B is on the MetaFloor roadmap. This page reserves its home in the Manifest suite β weights, benchmarks, and usage will be published here at release.
Manifest is MetaFloor's suite of supply-chain expert models β purpose-built specialists in procurement, demand planning, warehouse operations, supplier relationship management, risk & resilience, transportation, inventory, and order fulfilment.
Manifest 30B is the frontier of the suite β designed for enterprise-scale supply-chain intelligence and the deepest, most nuanced reasoning MetaFloor will offer.
The Manifest family
Two kinds of models:
π§ Orchestrators β general-purpose, handle any supply-chain area
| Model | Size | Preferred over base | Status |
|---|---|---|---|
| Manifest 0.8B | 0.8B | 68.5% | β available |
| Manifest 2B | 2B | 88.4% | β available |
| Manifest 4B | 4B | 90.9% | β available |
| Manifest 30B | 30B | β | π coming soon |
π― Domain-experts β specialized for a single area
| Model | Preferred over base | Status |
|---|---|---|
| Manifest Specialist Β· Risk & Resilience | 72.5% | β available |
| Manifest Specialist Β· Inventory Optimization | 77.5% | β available |
| Manifest Specialist Β· Demand Planning | 82.5% | β available |
Orchestrators are scored on the general supply-chain benchmark; domain-experts on their focused domain benchmark.
Not just a model β a full stack
Manifest is backed by everything needed to build it and trust it:
- A purpose-built dataset β thousands of supply-chain instructionβresponse pairs spanning 8 sub-domains and every company scale, generated by a seed-driven operator-as-teacher pipeline.
- A reproducible training pipeline β documented LoRA fine-tuning.
- An independent benchmark β 116 held-out expert questions, scored blind by a panel of LLM judges.
We built the model, the data, and the evaluation.
Available now
While Manifest 30B is in the works, these are live today:
- Manifest 0.8B β fast, low-cost, high-volume
- Manifest 2B β the balanced everyday default
- Manifest 4B β the most capable available today
Want early access to Manifest 30B? Reach out to MetaFloor.
License
Manifest models and the supply-chain-eval benchmark are released under CC-BY-NC-4.0 β free for research and non-commercial use, with attribution. Commercial use requires a license from MetaFloor β get in touch at metafloor.ai.
- Downloads last month
- -