Hunyuan-T1
Hybrid Transformer-Mamba-MoE reasoning model, API-only
About
Hunyuan-T1 is Tencent's flagship reasoning model, notable as a large-scale hybrid of Transformer, Mamba, and mixture-of-experts architecture — Tencent says the Mamba components roughly double decoding speed versus a pure Transformer. Tencent reports 87.2 on MMLU-Pro and 96.2 on MATH-500, with particular strength in Chinese-language tasks, and prices API access aggressively at a Tencent-reported $0.14 per million input tokens. It is API-only: no open weights, hosted on Tencent Cloud with integration into Tencent Docs and the Yuanbao chatbot. That makes it easy to try but ties you to Tencent's platform and its data-residency posture.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Tencent-reported: MMLU-Pro 87.2, MATH-500 96.2, GPQA Diamond 69.3, LiveCodeBench 64.9, AIME 2024 78.2
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (verified tier, metrics-v1). Evidence: Tencent release March 21 2025 confirmed across outlets: hybrid Transformer-Mamba-MoE, 128K context.; Benchmarks Tencent-reported: MMLU-Pro 87.2, MATH-500 96.2, GPQA Diamond 69.3, LiveCodeBench 64.9.; Pricing $0.14/1M input tokens is Tencent-reported.; Limitation: API-only, no open weights; tied to Tencent Cloud..