HASTATAgent stack guide mercury
Mercury Coder Small
Model by Inception AI. A diffusion large language model that runs incredibly quickly (500+ tokens/second) while matching Claude 3.5 Haiku and GPT-4o-mini. 1st in speed on Copilot arena, and matching 2nd in quality.
Fresh · fetched 2026-08-17T02:11:57.739Z · Models.dev(opens in a new tab)
Minimum context32K
Output16K
Providers2
CapabilitiesTool calling
Provider offers
| Providers | Input / 1M | Output / 1M | Cache read / 1M | Context / output | Updated |
|---|---|---|---|---|---|
 NanoGPTmercury-coder-small | $0.25 | $1.00 | $0.13 | 32K | 2024-01-01 |
 Vercel AI Gatewayinception/mercury-coder-small | $0.25 | $1.00 | Unavailable | 32K | 2025-02-26 |
Lowest reported provider price in USD per 1M tokens; unavailable is not free. Provider documentation remains authoritative. Catalogue presence is not an uptime, security, or quality endorsement.
Source metadata and limitations
Capabilities are source-reported. Tool calling does not prove quality in a particular harness. Prices and limits can change.
Modalities: text → text