Tencent: Hy-MT2-1.8B

Tencent Hy-MT2 1.8B sits within the broader Hy-MT2 lineup, which the vendor released in both 1.8B and 7B parameter sizes to address machine translation workloads. The 1.8B variant is positioned as the lightweight option of the pair, aimed at situations where a smaller footprint is preferable, while still drawing on the same underlying Hy-MT2 training approach. The family is shipped under the Tencent HY Community License Agreement, which formally governs redistribution and hosted service use of the Tencent HY Works and Documentation, with the agreement explicitly tied to a Hy-MT2 release milestone in 2026.

The model's intended use is translation, and the community packaging around it emphasizes an Ollama-compatible prompt template designed to keep local translation behavior consistent across runs. Because the same Hy-MT2 family also has a larger 7B sibling, the 1.8B version is best understood as a speed- and resource-oriented choice for translation pipelines that do not need the heavier model's capacity. Practical fit therefore centers on cost- or latency-sensitive translation scenarios, while users who want broader headroom can step up within the same Hy-MT2 family rather than switching frameworks.