Inception has launched Mercury 3, its fastest and most capable diffusion-based large language model to date, achieving 1,068 tokens per second and outperforming leading competitors.
TL;DR
- Inception's Mercury 3 sets new benchmarks in speed and performance, running 4-8x faster than competitors.
- Early-access partners report significant improvements in latency-sensitive applications, with Mercury 3 completing 50% of Mind2Web tasks.
- Mercury 3 is part of Inception's real-time AI portfolio, which includes Mercury Voice and Mercury Decide, and is available through multiple API platforms.
What happened
Inception introduced Mercury 3 on October 19, 2026, claiming it is the most capable diffusion-based large language model released to date. The model is priced at $0.10 per million input tokens and $0.35 per million output tokens, with a 50% discount applied at launch. Cached input tokens are priced 90% below the input rate, at $0.02 per million tokens and $0.01 under the promotion.
Mercury 3 outperforms GPT-6 Luna and Claude Haiku 5.5 across a range of benchmarks while running four to eight times faster at more than 1,000 tokens per second. The model achieves a speed of 1,068 tokens per second on NVIDIA GPUs and has a context window of one million tokens, up from 260,000 tokens for Mercury 2.5.
Early-access partners are already using Mercury 3 in latency-sensitive applications, including browser-use agents, interactive education, real-time search agents, and customer support. The company reported that Mercury 3 completed 50% of 50 live Mind2Web navigation tasks, a completion count significantly higher than GPT-6.1-Sol's, achieved with the same token volume and the same number of agent steps at a fraction of the per-token cost.
Why it matters
Mercury 3's performance and speed make it a strong competitor in the AI market, particularly for applications requiring real-time processing. Its ability to outperform leading models like GPT-6 Luna and Claude Haiku 5.5 positions it as a viable option for developers and startups looking for high-performance AI solutions.
The model's success in early-access production results demonstrates its potential for real-world applications. Partners like TinyFish AI, General Learning, and Octen AI have reported significant improvements in latency and performance, highlighting Mercury 3's versatility and efficiency.
Inception's broader Mercury portfolio, including Mercury Voice and Mercury Decide, offers a range of solutions for different AI applications. The availability of Mercury 3 through multiple API platforms, including OpenRouter, Baseten, and Vercel, makes it accessible to a wide range of users and enterprises.
Key facts
- Mercury 3 is priced at $0.10 per million input tokens and $0.35 per million output tokens, with a 50% discount applied at launch.
- The model achieves a speed of 1,068 tokens per second on NVIDIA GPUs and has a context window of one million tokens.
- Mercury 3 outperforms GPT-6 Luna and Claude Haiku 5.5 across a range of benchmarks.
- Early-access partners report significant improvements in latency-sensitive applications, with Mercury 3 completing 50% of 50 live Mind2Web navigation tasks.
- Mercury 3 is available through the Inception API Platform, OpenRouter, Baseten, and Vercel, and is OpenAI API compatible.
- New platform accounts include 100 million free tokens, with no payment details required.
- Mercury 2 and Mercury 2.5 are no longer available to new users, with existing customers retaining access until November 15, 2026.
Context
Inception's release of Mercury 3 comes at a time when the demand for high-performance, real-time AI solutions is growing. The model's ability to outperform leading competitors and its success in early-access production results position it as a strong contender in the AI market.
The broader Mercury portfolio, including Mercury Voice and Mercury Decide, offers a range of solutions for different AI applications. This diversity makes Inception a versatile player in the AI industry, capable of addressing a wide range of use cases.
The availability of Mercury 3 through multiple API platforms ensures that developers and startups have easy access to this powerful AI model. This accessibility is crucial for fostering innovation and driving the adoption of advanced AI technologies.