Inception Releases Mercury 3 Diffusion LLM at $0.20 per Million Input Tokens

Inception has released Mercury 3, which the company's blog calls its most capable diffusion-based large language model to date. The blog post is dated October 19, 2026. OpenRouter's model page for inception/mercury-3 lists a release date of October 16, 2026.

Price, speed and context

Inception's models page lists Mercury 3 at $0.20 per million input tokens and $0.70 per million output tokens. At launch, Inception says Mercury 3 is 50% off, at $0.10 per million input tokens and $0.35 per million output tokens. Cached input is listed at $0.02 per million tokens at the standard rate and $0.01 per million tokens at the launch rate. The blog says cached tokens are 90% off the input price.

Inception says Mercury 3 runs at 1,068 tokens per second on NVIDIA GPUs and has a 1 million token context window, up from 260,000 for Mercury 2.5.

How Inception compares Mercury 3

Inception says that across a range of benchmarks, Mercury 3 outperforms GPT-6 Luna and Claude Haiku 5.5, while running 4 to 8 times faster. In a glance section, Inception also describes the model's quality as on par with or better than frontier speed-optimized models such as GPT-6 Luna and Claude Haiku 5.5. The blog post does not list the individual benchmark scores next to those claims.

Where Mercury 3 is available

Inception says Mercury 3 is available on the Inception API Platform and through OpenRouter, Baseten and Vercel, and that the API is OpenAI compatible. On OpenRouter, the model ID is inception/mercury-3. On the Inception API, the models page example uses the name mercury-3.

OpenRouter posted on X at 00:54 Beijing time on October 20 that Mercury 3 is live on OpenRouter, citing 1,000+ tokens per second, 1 million context and the same launch price of $0.10 per million input and $0.35 per million output.

Mercury 2 and Mercury 2.5

Inception's models page says Mercury 2 and Mercury 2.5 are no longer available to new users. Existing customers with usage before October 19, 2026 keep access until November 15, 2026. The page says enterprise customers remain supported under existing agreements and points to migration guidance for Mercury 3.