Open Model Ecosystem
Flagship families, from
enterprise to edge.
We prioritize models that offer high performance-per-parameter, native multimodality, and efficient deployment anchors for real-world products.
Model Recommender
Write a use case and get a ranked model shortlist.
Score 1 means strongest fit, 2 means a middle match, 3 means a weaker match, and 4 means the weakest acceptable match.
Example use cases
How it works
Try a real use case
Describe the job, the input format, deployment limits, privacy needs, languages, and whether the model must run locally or in the browser.
These cards come from matched Supabase examples. 1 is the strongest match and 4 is the weakest acceptable match.
IBM
Granite 4.0
Enterprise-grade agents and edge devices. A strong case study for hybrid Mamba-Transformer (9:1) architectures that optimize performance and memory.
- Hybrid MoE and dense models for extreme efficiency
- Native support for Document Vision and Safety guardrails
- Sovereign deployment for private enterprise data
Gemma 4
Agents and edge devices. A state-of-the-art family for multimodal reasoning with native audio and video processing across all sizes.
- Native Audio input on E2B and E4B edge models
- Mixture of Experts (MoE) for high-throughput cloud logic
- Effective 2B/4B footprints for mobile and IoT
IBM
Granite Time Series
IBM's specialized time-series family for forecasting, anomaly detection, representation learning, and similarity search. Built to stay compact, practical, and strong on structured temporal data.
- Purpose-built foundation models for forecasting and time-series understanding
- Tiny footprints that stay practical on CPU and constrained inference paths
- Coverage from zero-shot forecasting to anomaly detection and retrieval-style tasks
Alibaba
Qwen 3.6
A flexible family for agentic reasoning that stays practical at smaller sizes. Best understood through the 27B, 9B, 4B/2B, and 0.8B variants that span servers, workstations, and local devices.
- Qwen 3.6-27B as the strong mid-to-large reasoning anchor
- Qwen 3.6-9B and 4B/2B for workstation and on-device agent flows
- Qwen 3.6-0.8B for tiny command, routing, and control tasks
Mistral AI
Mistral Family
The premier European AI provider. A practical lineup spanning compact edge models to frontier enterprise reasoning.
- Ministral 8B for dense edge deployment
- Mistral Small and Medium for balanced product reasoning
- Mistral Large for top-tier enterprise capability
Decision Framework
How to choose your anchor.
Picking a model family is a decision about hardware, latency, and reasoning depth. Below is how we frame the current landscape of flagship open systems.
Granite 4
Best framed as the enterprise-grade agent and edge device family for trusted, domain-focused corporate deployment.
Gemma 4
Best framed as the agent and edge device family for local-first intelligence and high capability density on constrained hardware.
Granite Time Series
Best framed as the specialized family for forecasting, anomaly detection, and representation learning on structured temporal data.
Qwen 3.6
Best framed as the specialized family for strong agent-related tasks and complex tool-use reasoning across many modalities.
Mistral Family
Best framed as the premier European AI provider for versatile agents and efficient edge-native deployments.