Products/AI/LLM Infrastructure/Mercury Edit 2

Mercury Edit 2

A new frontier in LLM speed

AI/LLM InfrastructureaillminfrastructurespeeddiffusionReviewed
Mercury Edit 2

Our Take

Mercury Edit 2 is doing next-edit prediction rather than the overdone next-token completion thing, which is actually a smarter angle — predicting the actual change you want to make means less back-and-forth between you and the model. That 159 votes on what I'm guessing is a Product Hunt launch tells me real devs are at least curious, though I'd want to know actual latency numbers before hyping it up. The move here is precision over speed, but whether that's the right bet for your workflow depends on what you're building.

Builds and deploys next-generation large language models (LLMs) powered by diffusion rather than traditional autoregressive generation. Their models produce many tokens in parallel, making them several times faster and less than half the cost of conventional LLMs.

Key Features
Diffusion-based language generation (not autoregressive), Parallel token generation - generates tokens simultaneously, 5x greater speeds than conventional LLMs, Less than half the cost of traditional models, Fine-grained control over outputs, Schema adherence capabilities, Unified paradigm for combining language with audio, images, and video
Problem It Solves
Solves the speed and cost constraints of traditional autoregressive LLMs that generate one token at a time sequentially, which makes latency and cost scale with every additional token.
Target Customer
Fortune 500 companies deploying AI in production applications
Use Cases
Real-time voice interactions, Instant AI agents for automation, Code editing with low latency, Fast co-pilots for creative work, Rapid search from knowledge bases
Pricing Details
At a fraction of the cost of other top-tier models
Differentiator
Diffusion-based approach instead of autoregressive - generates tokens in parallel rather than sequentially
Why Now
As AI systems move from single-shot prompts to multi-step agents, output speed becomes a first-order constraint. The more the model thinks at inference time, the more important it is that it can respond fast enough to ship in real products.
Traction
Customers Mentioned: Fortune 500 companies · Notable Metrics: 5x greater speeds with best-in-class quality

Key Facts

Category
AI/LLM Infrastructure
Pricing
Input: $0.25 per 1M tokens; Output: $0.75 per 1M tokens
Discovered via
product-hunt

The people behind Mercury Edit 2

A

Aditya Grover

profile

Co-Founder

Inception Labs - UCLA professor, co-founded 2024.

S

Stefano Ermon

profile

Co-Founder

Inception Labs - Stanford professor, co-founded 2024. Diffusion-based LLM pioneer.

V

Volodymyr Kuleshov

profile

Co-Founder

Inception Labs - Cornell professor, co-founded 2024.

Links

Browse by category

Want products like this in your inbox every morning?

Five products. Every morning. Written by someone who actually cares whether they're good or not. Free forever, unsubscribe whenever.

Mercury Edit 2 — SLAYREPORT