> ## Documentation Index
> Fetch the complete documentation index at: https://docs.portkey.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# June

June brings deeper MCP capabilities and richer observability across the gateway.

We've also shipped new guardrail partner, expanded provider support, and improvements across Anthropic, Bedrock, and Vertex AI.

See what's new:

## Summary

| Area | Key highlights |
| :- | :- |
| **Gateway** | • Server-Side MCP Execution (Beta) <br /> • OTel guardrail spans <br /> • Token metering for MCP and agents <br /> • MCP OTel log export |
| **Guardrails** | • Alibaba Cloud AI |
| **Models and providers** | • Azure AI Inference: MAI image generation <br /> • MiniMax <br /> • Anthropic context management <br /> • Claude Code OAuth |

## Gateway

### Server-Side MCP Execution (Beta)

MCP tools now work through the AI Gateway on **any provider**—including AWS Bedrock and Vertex AI, which don't natively support remote MCP. The gateway handles fetching tool definitions and executing calls server-side, so you get MCP capabilities without any provider-level changes.

### OTel: Guardrail execution spans

Guardrail checks now emit their own OTel spans, so exactly what your guardrails did right alongside the model call in your traces.

### Token metering for MCP and agent requests

Every token from **MCP requests** and **agent endpoints** now shows up in your analytics and logs alongside LLM usage—one complete picture of consumption across all your AI workloads.

### MCP gateway: OTel log export

MCP traffic now appears in the same observability pipeline as your LLM workload exported to your OpenTelemetry collector alongside gateway logs.

## Guardrails

### Alibaba Cloud AI

**Alibaba Cloud AI** is now available as a guardrail for content safety checks, configurable through the standard guardrails workflow.

## Models and providers

### Azure AI Inference

* Image generation and editing now supported for MAI models (e.g., MAI-Image-2.5)

### New providers

* **MiniMax** — chat completions

### Anthropic

* **Context management:** `context_management` passed through to the upstream API, enabling memory tools and managed-context workflows on Claude models
* **Claude Code OAuth:** Requests from Claude Code using an OAuth bearer token now forward `Authorization` to Anthropic correctly
* **File downloads:** Binary responses (e.g., PPTX from Skills + Code Execution) now pass through untouched
* **Thinking blocks in streaming:** Reasoning summary blocks are now included in streaming Responses API responses for Anthropic models

## Bug fixes and improvements

* Vertex AI: MoonshotAI model provider now supported
* Modal: embeddings now supported alongside chat completions
* Fireworks AI: `stream_options` support added
* Amazon Bedrock Guardrails now only block when the action is blocked

## Support

<CardGroup cols={2}>
  <Card title="Need Help?" icon="bug" href="https://github.com/Portkey-AI/gateway/issues">
    Open an issue on GitHub
  </Card>

  <Card title="Join Us" icon="discord" href="https://portkey.wiki/community">
    Get support in our Discord
  </Card>
</CardGroup>


## Related topics

- [June](/docs/changelog/2025/june.md)
- [Enforcing Saved-Only Resources](/docs/product/administration/enforce-saved-only-config.md)
- [Backend](/docs/changelog/backend.md)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.