Https Www Tryproxy Io Blog Proxyai Inception

ProxyAI - Blog

**Blog**

Product updates, engineering notes, and announcements from ProxyAI.

Today we are partnering with Inception to bring faster, more consistent code edits to ProxyAI.

Meet Mercury Coder

Mercury Coder from Inception is a diffusion Large Language Model (dLLM). Unlike standard autoregressive LLMs that generate output token by token, a dLLM refines drafts across many spans at once. This makes it especially good at consistent multi line edits and structural code changes.

ChatTab

Your browser does not support the video tag.

Why We Switched

Hosting and tuning large models in house is expensive and operationally heavy. Mercury Coder is tuned for the tasks we already provide out of the box, so we are leaning on a specialized partner for model training and serving while we focus on product quality and developer workflow.

What You Will Notice

You will use the same features as before, now powered by Mercury Coder. Here is what improves:

• Autocomplete: faster completions in the editor with improved accuracy in context.
• Next-edit suggestions: more consistent multi line edits and better structural updates across related code.
• Auto apply: quicker application of changes, up to 8k tokens per second depending on context.

Availability

Live today. Mercury Coder is the default model in ProxyAI for autocomplete, next edit, and auto apply, delivering faster and more accurate multi line edits. Not using our cloud offering? Connect your Inception API key to use the same features. Update to the latest ProxyAI and try it in your current flows.

Read the provider guide:Inception on ProxyAI

We use cookies to ensure that we give you the best experience on our website. Read cookies policies.

Reject AllAccept All Cookies