You control which LLM provider families and model IDs are available in your organisation. Fontana ships 300+ supported models across leading providers: frontier chat and reasoning models, open-weight endpoints, and domain-specific hosts, all routed through your approved gateways and BYOK keys (see Gateways
When you add or choose an OpenRouter model, the model list header has a nitro toggle (on / off ). That header is the agent model dialog, and AI → Gateways and Models while OpenRouter is open. An approved OpenRouter model has the same toggle on its card. The model list shows the model name and a nitro badge. Turning nitro on sends that model to the fastest provider. A priority endpoint is used only when it is the fastest, and that request is billed at the priority rate. Other gateways stay unchanged. See OpenRouter Nitro .
Icon Provider Examples Documentation … OpenAI GPT, o-series, and reasoning models routed through your approved gateways. platform.openai.com … Anthropic Claude family models for analysis, drafting, and review inside governed workflows. docs.anthropic.com … Google Gemini Gemini and other Google Generative AI models under your credential and policy controls. ai.google.dev … Meta Llama Open-weight Llama models from Meta available through configured gateway routes. llama.com … Mistral Mistral Large, Small, Pixtral, and Mixtral models from direct or aggregated gateways. docs.mistral.ai … DeepSeek DeepSeek reasoning and coding models when your gateway catalogue includes them. api-docs.deepseek.com … Cohere Command and enterprise Cohere models for retrieval-heavy operational work. docs.cohere.com … xAI Grok Grok models available when your approved gateway mix includes xAI routes. docs.x.ai … NVIDIA Nemotron and hosted LLMs via NVIDIA NIM and compatible gateway paths. build.nvidia.com … Qwen Alibaba Qwen models for multilingual analysis and structured extraction tasks. qwen.ai … Perplexity Perplexity-hosted models when your procurement approves those gateway routes. docs.perplexity.ai … Microsoft Azure Azure OpenAI and Phi models under Microsoft agreements your team already holds. learn.microsoft.com … AWS Titan, Nova, and other AWS-hosted foundation models via Bedrock-compatible routes. docs.aws.amazon.com/bedrock … AI21 Jurassic and Jamba models for teams that standardise on AI21 through approved gateways. docs.ai21.com … Yi 01.AI Yi models for cost-efficient inference on supported gateway catalogues. lingyiwanwu.com … TII Falcon Falcon open models from the Technology Innovation Institute on supported routes. falconllm.tii.ae … Nous Research Fine-tuned and Hermes model lines from Nous Research where gateways expose them. nousresearch.com … Inflection Inflection-hosted models when your gateway configuration includes those endpoints. inflection.ai … Liquid Liquid AI models for efficient inference on gateways that list them in catalogue. liquid.ai … 300+ models Frontier, open-weight, and domain-specific models across every major provider. openrouter.ai/models
If your preferred provider or model is not listed above, or you run internal inference endpoints, contact the Fontana Engineering team. We can add support for custom gateways, private hosts, or enterprise-only model access to match your deployment boundary.
