Hy-MT2-30B-A3B is Tencent's flagship translation model in the Hy-MT2 family — a 30B-parameter Mixture-of-Experts model with roughly 3B active parameters per token (the "A3B" suffix). It supports 33 language pairs plus five Chinese dialect and minority-language pairs, with workflows for structured, delimiter-based, contextual, and glossary-based translation.
| Developer | Tencent |
|---|---|
| Model ID | tencent/hy-mt2-30b-a3b |
| Release Date | August 20, 2026 |
| Context Window | 8,192 tokens |
| Max Output Tokens | 4,096 |
| Architecture | Mixture-of-Experts, 30B total / ~3B active parameters |
| Modalities | Text (input) → Text (output) |
| Capabilities | Streaming, structured (JSON schema) output |
As the largest model in the Hy-MT2 family, it targets the highest translation accuracy across the family's supported language pairs.
Only a fraction of its 30B parameters activate per token, aiming to keep inference costs closer to a smaller dense model while retaining a larger model's capacity.
Structured, delimiter-based, contextual, and glossary-based translation modes support both raw and terminology-constrained translation.
The largest model in the family, suited to translation work where quality matters more than raw throughput cost.
Glossary-based and contextual workflows help keep terminology and meaning consistent in longer or domain-specific text.
Covers five Chinese dialect and minority-language pairs in addition to the standard 33.
Measured on the AI/ML API gateway across live traffic. Figures are medians unless stated. Last updated 2026-10-12.
Model-specific sources describe Hy-MT2-30B-A3B as a multilingual translation model suited to general translation tasks across 33 languages, including instruction-following translation in multiple languages.
It is also documented for real-world business translation, where it is positioned for complex practical scenarios and broad business use.
The model is further described as suitable for domain-specific translation, with reported strength in professional domains such as finance, law, and politics.
Sources also characterize it as useful for structured, glossary-based, context-aware, and style-guided translation workflows.
import requests r = requests.post( "https://api.aimlapi.com/v1/chat/completions", headers={"Authorization": "Bearer " + AIMLAPI_KEY}, json={ "model": "tencent/hy-mt2-30b-a3b", "messages": [ { "role": "user", "content": "Hello!" } ] }, ) print(r.json())
const r = await fetch("https://api.aimlapi.com/v1/chat/completions", { method: "POST", headers: { Authorization: `Bearer ${process.env.AIMLAPI_KEY}`, "Content-Type": "application/json", }, body: JSON.stringify({ "model": "tencent/hy-mt2-30b-a3b", "messages": [ { "role": "user", "content": "Hello!" } ] }), }); console.log(await r.json());
curl -X POST https://api.aimlapi.com/v1/chat/completions \ -H "Authorization: Bearer $AIMLAPI_KEY" \ -H "Content-Type: application/json" \ -d '{"model":"tencent/hy-mt2-30b-a3b","messages":[{"role":"user","content":"Hello!"}]}'
OpenAI-compatible — swap the base URL and it works with your existing SDK.
| Type | Price |
|---|---|
| Input | |
| Output | |
| Model | Input | Output | Context | Best for |
|---|---|---|---|---|
Hy-MT2-30B-A3B This page | High-accuracy translation | |||
| Reasoning + agents | ||||
| Balanced coding + agents | ||||
| Long-context, multimodal & agentic workflows | ||||
| Reasoning + agents |
Hy-MT2-30B-A3B has a 8,192 tokens context window and can return up to 4,096 tokens.
Hy-MT2-30B-A3B takes text as input and returns text.
Use tencent/hy-mt2-30b-a3b as the model id. Requests go to https://api.aimlapi.com/v1/chat/completions.
Hy-MT2-30B-A3B became available on August 20, 2026.
Hy-MT2-30B-A3B is priced at input $0.1017796 / 1M tokens, output $0.405743 / 1M tokens.
Yes, Hy-MT2-30B-A3B can stream responses as they are generated.
Hy-MT2-30B-A3B was built by Tencent.
Tool calling is not listed among its capabilities, which include streaming and structured output.
