Verified, then logged.
Sun 22 Nov 2026 · 00:34 KST / 15:34 UTC한국어 | ENSearch /

Breaking01 / AI

Inception prices Mercury 3 at $0.20 per million input tokens

A rack server pulled open in a data center aisle reveals tightly packed GPU accelerator cards, with a technician walking away out of focus in the background. AI-generated image

Inception priced Mercury 3, released on 19 Oct 2026, at $0.20 per million input tokens and $0.70 per million output tokens, the company said.

Pricing and access. The launch discount is 50% ($0.20 to $0.10 per million input tokens, $0.70 to $0.35 per million output tokens), according to the announcement. Cached tokens are 90% off the input price. The model is available on Inception's API platform, OpenRouter, Baseten and Vercel, and is compatible with the OpenAI API.

Mercury 3 is a diffusion language model. Diffusion is a method that generates text by refining a noisy draft over several passes instead of producing one token at a time. Inception published a speed of 1,068 tokens per second on NVIDIA GPUs, and context, the span of tokens a model handles at one time, grew from 260K tokens on Mercury 2.5 to 1M tokens.

Inception wrote in its announcement that "Mercury 3 demonstrates that text diffusion scales." Total parameters are more than 40 times those of Mercury 1. The benchmark intelligence measure is three times that of Mercury 1 (mini), a company-published figure.

Inception named GPT-6 Luna and Claude Haiku 5.5 as its comparison models. By its own published figures, Mercury 3 beats both on several benchmarks and runs 4 to 8 times faster.

TinyFish AI, an early-access partner, tested Mercury 3 on 50 Mind2Web web navigation tasks. Shuhao Zhang, the firm's co-founder, wrote in Inception's announcement that the model completed 50% of them, more than GPT-6.1-Sol (a company-published figure).

Inception has not disclosed when the launch discount ends.