Teams building AI agents want to use the best model for each job.
The catch? The best model keeps changing, sometimes week to week.
Switching models shouldn’t mean new integration work or another security review. That’s why broad model support matters so much to us at MuleSoft: our customers should be able to use the right model as soon as it arrives, keeping pace with their AI innovation.
Which is why we’re excited to announce that Omni Gateway now supports more than 140 large language models from OpenAI, Anthropic, Google, Amazon Bedrock, Microsoft Azure, Mistral, xAI and NVIDIA. That includes the latest frontier models: Claude Opus 5, GPT-6 Astra, Gemini 3.8 Flash and Grok 4.6.
Every model sits behind the same gateway that already governs your APIs, MCP servers and agents, and that gateway is included in your MuleSoft subscription.
Developers send requests to a single endpoint, platform teams decide which models are behind it, and every request is governed the same way.
Model choice without the rewiring
Each new model release usually means new integration work: another API key, another endpoint, another set of rules to configure. With Omni Gateway, adopting a new model changes the routing, not the application. You can move a workload from Claude Sonnet 4.6 to Claude Sonnet 5, or from GPT-5.4 to GPT-6 Astra, without touching the agents and apps that call it.
A single model end point can also route to several models. It chooses by the model named in the request, the intent of the prompt, or request headers, so one endpoint can serve a whole family of models.
Use the clouds you already pay for
Many enterprises buy models through their cloud provider. Omni Gateway supports models hosted on Amazon Bedrock and Microsoft Azure alongside direct provider APIs. You can use your existing cloud commitments and still keep a single governance layer across all of them. Support for Microsoft Foundry and more Google-hosted models is coming soon.
New model inherit your existing policies
A model added behind Omni Gateway inherits the policies you already enforce: the same authentication, rate limits, threat protection and audit trail. Security teams can approve a newer model without having to write new policies for it. .
Govern LLM traffic alongside your agents, MCP tools and APIs
Most AI gateways govern LLM calls in isolation. Omni Gateway is part of Agent Fabric, so model traffic is governed in the same place as your agents, MCP tools and APIs, with one set of policies across all of them. The user’s identity travels with each request through your identity providers such as Okta or Entra. Policies can be enforced across the gateways you already run, including Kong, Apigee and Azure. If an agent misbehaves, Kill Switch lets you quarantine it before it causes further harm.
Spend where it counts
Not every request needs a frontier model. Omni Gateway lets you intelligently route simple tasks to lower-cost, faster models and track spend by team. Cost optimization strategies such as semantic caching lets Omni Gateway reuse responses for prompts that mean the same thing, even when the wording differs, so you don’t pay for the same answer twice. Keep the most capable models for the hardest problems, and set usage limits so spend stays predictable.
What you can build
With expanded model support, teams can:
- Give developers a single, approved way into AI. One endpoint and one set of credentials covers every model the organization has approved.
- Test new models safely. Send part of your traffic to a newly released model, compare results, and roll it out when it’s ready.
- Run multi-model agents. Use a reasoning model for planning, a fast model for classification and a code model for generation, all governed together.
- Stay portable. Avoid lock-in to a single provider, and switch quickly when pricing, performance or policy changes.
Bring your own
If the model you need isn’t in the catalog, bring it yourself. Any model that uses the OpenAI, Anthropic or Gemini API format can be added by publishing custom LLM metadata.
Availability
Expanded model support is available now in Omni Gateway. Managed Omni Gateway is now included with your existing MuleSoft subscription.
Model choice is only one part of running agents in production. To see everything Omni Gateway has to offer, watch our overview video. For regular updates on integration, APIs, and AI, subscribe to Technically Speaking, MuleSoft’s newsletter on LinkedIn.
