Hy-MT2 30B-A3B

by Tencent · 30B (A3B) parameters

Current budget open weights cheap tier

Email me when Hy-MT2 30B-A3B gets cheaper

Today's price · per 1M tokens

Input

$0.074

per 1M tokens

Output

$0.295

per 1M tokens

Blended

$0.129

blended $/1M — 3:1 weighted input:output

How this price is scoped: Tencent's own price, quoted by the FP8 endpoint Tencent operates for this model. Tencent's RMB rate card carries the matching row at ¥0.5 in / ¥2 out per 1M tokens, which converts to these figures at 6.78 CNY/USD. That card names service tiers (Hy-MT2-Pro, Hy-MT2-Plus, Hy-MT2-Lite) rather than weight classes, and Pro and Plus share the same ¥0.5/¥2 row, so which tier this model is sold under is not stated on any Tencent page and is not claimed here. Tencent publishes no cached-input rate for this family.

Source: official Tencent pricing · read Aug 20, 2026 MODELPRICEWATCH.COM · 2026-08-20

Price receipt

No confirmed receipt. We hold dated captures of Tencent's pricing page, but none of them tied Hy-MT2 30B-A3B to a published price on that page, so we do not claim this number is sourced. Check the provider's page directly before relying on it.

Overview

Tencent Hy-MT2 30B-A3B — the largest model in Tencent's Hy-MT2 "fast-thinking" translation family, a 30B mixture-of-experts with roughly 3B active parameters. It translates among 33 languages and five Chinese dialect and minority-language pairs, and handles structured, delimiter-based, contextual, glossary-based and style-guided translation. Tencent open-sourced the weights on 2026-05-21 and also sells the family per token on its Tencent Cloud TokenHub rate card, serving this model itself as an FP8 endpoint. Context window 8,192 tokens with a 4,096-token maximum output, per Tencent's own model list — this is a translation model, not a long-context general assistant.

Run it yourself