Mercury 3 now available on AI Gateway
- Source
- vercel.com
- Author
- Jerilyn Zheng
- Date
Context
Vercel's AI Gateway has added Mercury 3 from Inception at half its list price for launch, and the reason to care is speed (vercel.com). Most language models write left to right, one at a time, so each token waits for the one before it. Mercury is a diffusion model: it starts from a rough draft of the output and refines it over several steps, changing many tokens at once (Inception). Inception reports generation above 1,000 tokens per second, and its own announcement gives 1,068 tokens per second on NVIDIA GPUs. The changelog pitches it at latency-sensitive work such as browser agents, real-time search, interactive education and customer support, and gives it a 1M-token , up from 260K for Mercury 2.5 (Inception).
The changelog does not list prices. Inception lists $0.20 per million input tokens and $0.70 per million output, cut to $0.10 and $0.35 at launch, without stating when the discount ends; the gateway passes provider pricing through with no markup. On quality, the changelog says only that Inception reports strong results across a range of benchmarks. Inception's announcement offers one concrete test: in its own production web-, Mercury 3 completed 50% of 50 live Mind2Web navigation tasks, which it calls significantly more than GPT-6.1-Sol. These are vendor-run results.
- token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
- context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
- agent harness — The scaffolding around a model that turns it into a working agent — the loop, the tools it can call, and the rules for when to stop.
- LLM — A large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.
Checking sign-in…
Loading comments…