Foundation models
General-purpose large language and multimodal models.
7 tools. Adoption level is the effort to run it well, not a quality score. Limitations are sourced; a blank means none documented yet.
| Tool | Best for | Level | Pricing | Key documented limitation |
|---|---|---|---|---|
| Claude | Reasoning, long-doc analysis, structured extraction. | L2 Assisted | $ | None documented yet |
| Cohere | Enterprises requiring sovereign AI and strictly private model deployments. | L3 Integrated | Free trial + Usage-based + Enterprise cloud partners | None documented yet |
| Gemini | Multimodal + long-context tasks. | L2 Assisted | $ | None documented yet |
| Gemma | Teams that need a local fallback when hosted LLM providers degrade or for privacy-sensitive workloads. | L4 Orchestrated | Free / Open Weights | None documented yet |
| Llama | Teams running their own inference for cost control, data residency, or provider redundancy. | L4 Orchestrated | Free / Open Weights | None documented yet |
| Mistral AI | Research and development teams working with state-of-the-art open models. | L2 Assisted | Usage-based API + Enterprise (Mistral Large/Medium access via La Plateforme) | None documented yet |
| OpenAI | General-purpose AI in apps and agents. | L2 Assisted | $ | None documented yet |