Tencent

Gander

Tencent

Gander is a 9B full-duplex audio-visual interaction model developed by Tencent’s Hunyuan Speech team with academic collaborators. Its MiniCPM-o 4.5-derived Thinker and streaming Talker handle speech, images, video frames, text, turn taking, interruption, and structured delegation while tasks continue asynchronously. Apache-licensed model components and a serving runtime are publicly available. The larger system pairs this conversational model with a separately configured reasoning agent for long-horizon execution; those back-end model capabilities are not assumed to reside inside the 9B checkpoint. The release date follows the September 8 technical report that publishes the named model; the project dates its checkpoint release to September 9.

Hy4 Preview

Tencent

Hy4 Preview is Tencent’s open-weight productivity-focused language model with 770B total parameters, 49B active per token, and a 1M-token context window. It targets code planning, debugging, office workflows, financial analysis, and research tasks that require sustained tool interactions. Tencent released weights under Apache 2.0 and made the model available through its productivity applications and inference APIs. Its preview status distinguishes this release from future Hy4 checkpoints; planned follow-up models are not assumed available. Reported improvements in internal task evaluations are provider findings rather than independent measurements.

Hy3

Tencent

Hy3 is Tencent’s stable follow-up to Hy3 Preview, trained with higher-quality post-training data and feedback from deployed products. It retains a 295B-parameter sparse architecture with 21B active parameters and a 256K context. The model supports both fast responses and slower reasoning, with improvements to tool use, coding, and productivity workflows. Apache-licensed weights replace the preview’s custom community license. Official API access and integration into Tencent products provide hosted alternatives to private deployment.

Hy MT2

Tencent

Hy MT2 is Tencent’s second-generation multilingual translation language model family. It includes 1.8B and 7B dense models and a 30B sparse tier with approximately 3B active parameters. The models support 33 languages and five regional languages or dialects, with terminology constraints, contextual translation, and improved instruction following. Apache-licensed weights and a translation application enable local and hosted evaluation. Quantized mobile builds are variants of the released language checkpoints; they are not separate language-model identities.

Hy3 Preview

Tencent

Hy3 Preview is the first major model released on Tencent’s rebuilt Hunyuan training infrastructure. Its mixture-of-experts language backbone has 295B total parameters and 21B active per token, with additional multi-token-prediction parameters. The model supports 256K context and targets reasoning, coding, instruction following, and agent workflows. Public weights use the Tencent Hy Community License, which differs from the later Apache-licensed stable Hy3 release. Preview limitations and licensing should be evaluated before deployment.

Hunyuan 2.0 Instruct

Tencent

Hunyuan 2.0 Instruct is Tencent’s hosted general conversation and instruction-following tier, launched on November 11, 2025 according to the official Hunyuan changelog. Built on the Hunyuan 2.0 foundation that replaces TurboS, it targets multi-turn comprehension, long documents, creative writing, knowledge accuracy, coding, and reasoning. The API documents 128K maximum input and 16K maximum output. Its separate API identity and output policy distinguish it from Hunyuan 2.0 Thinking. No public weights, parameter count, or detailed architecture are established by the release sources.

Hunyuan 2.0 Thinking

Tencent

Hunyuan 2.0 Thinking is Tencent’s hosted reasoning model, publicly listed as launched on November 9, 2025 in the official Hunyuan product changelog. It replaces the TurboS foundation with Hunyuan 2.0 and emphasizes complex instruction following, multi-turn and long-document understanding, code, agents, and inference-intensive problem solving. The API offers 128K maximum input and 64K maximum output. It is a distinct named hosted reasoning tier from Hunyuan 2.0 Instruct, rather than an assumed downloadable checkpoint. Network details and parameter counts are omitted because the cited API release documentation does not establish them.

Hunyuan Large

Tencent

Hunyuan Large is Tencent’s open-weight 389B-parameter mixture-of-experts language model with 52B active parameters. The release publishes pretrained, instruction-tuned, and FP8 instruction checkpoints together with training and inference code. Grouped-query and cross-layer attention reduce key-value cache requirements for long documents. The pretrained model supports 256K context, while the instruction model uses 128K; this page records the conservative conversational limit. Its custom Tencent license governs downloadable weights, and local or Tencent Cloud TI deployments support customization. The November 18 instruction refresh is grouped as a variant.