Mercury Edit 2

Mercury Edit 2 is a purpose-built diffusion LLM designed to serve as a high-speed assistant for the most latency-sensitive aspects of software development. By utilizing diffusion to generate tokens in parallel, the model delivers next-edit predictions that integrate seamlessly into a developer's workflow, allowing for immediate acceptance of suggestions. It functions by analyzing a user's recent edits alongside the broader codebase context to anticipate and propose the next logical change, effectively acting as a responsive partner that keeps pace with the speed of human thought.

The model was developed using a carefully curated dataset of edits across diverse programming languages and scenarios to ensure high-quality, relevant suggestions. To refine its behavior and prevent overzealous or distracting edits, the training process incorporates a human preference dataset that tracks explicit feedback on accepted versus rejected suggestions. This data is leveraged through an unpaired reinforcement learning method known as KTO, which aligns the model's output with user preferences. This approach ensures that the model remains a practical, unobtrusive tool for developers, balancing predictive power with the precision required for real-world coding environments.

Inceptionmercury-edit-2mercury

Open provider page Provider docs

Quick Info

Provider: Inception
Model key: mercury-edit-2
Release date: Mar 30, 2026
Last updated: Mar 30, 2026
Input modalities
Output modalities
Capabilities

Cost

Input token cost: $0.25
Output token cost: $0.75

Limits

Output tokens: 8,192 tokens
Context window: 32,000 tokens

Latest news about Mercury Edit 2

Inception

Official sourceAnnouncement

The Inception blog index lists the Mercury Edit 2 launch post dated March 30, 2026, titled "Introducing Mercury Edit 2 — A next-edit diffusion LLM built for the fastest parts of your coding workflow." This places the Edit 2 announcement after the Mercury 2 launch on February 24, 2026, and before subsequent updates cove Other recent posts on the index cover Mercury 2 for Search (Aug 11, 2026), throughput improvements (Jul 29, 2026), Mercury 2 as a real-time phone-ready reasoning model (Jul 14, 2026), subagent workflows powered by Inception’s dLLMs at Augment Code (May 12, 2026), Azure Foundry availability (Jun 24, 2026), and PinchBenc

Inception

Official sourceAnnouncement

Inception introduced Mercury Edit 2, a diffusion-based large language model (dLLM) purpose-built for next-edit prediction in coding workflows, complementing its existing auto-complete endpoint. According to the official launch blog post by Kenan Hasanaliyev and Reece Shuttleworth, the model uses recent edit history and The model was trained on a curated dataset of edits across multiple languages, then aligned using an unpaired reinforcement learning method called KTO based on a human preference dataset of accepted versus rejected edits. Inception reports the new model’s edits are accepted 48% more often and are 27% more selective, re

Inception

CoverageComparison

What’s the difference between Claude Opus 4.6 and Mercury Edit 2? Compare Claude Opus 4.6 vs. Mercury Edit 2 in 2026 by cost, reviews, features, integrations, and more

Inception

CoverageComparison

What’s the difference between DeepSeek-V2 and Mercury Edit 2? Compare DeepSeek-V2 vs. Mercury Edit 2 in 2026 by cost, reviews, features, integrations, and more

Inception

CoverageComparison

What’s the difference between CodeGemma and Mercury Edit 2? Compare CodeGemma vs. Mercury Edit 2 in 2026 by cost, reviews, features, integrations, and more