Tencent Open-Sources Hunyuan-A13B: an 80B-Parameter MoE That Activates Only 13B
Tencent-Hunyuan · hf · 2026-09-24
Tencent Hunyuan released the technical report for Hunyuan-A13B, an open-source MoE LLM with 80B total parameters that activates only 13B at inference, balancing capability, efficiency, and deployment cost. It is pretrained on a rigorously filtered 20T-token corpus with enhanced STEM curation, then fine-tuned and aligned with large-scale RL.
- Dual-mode chain-of-thought: fast thinking for routine queries, slow thinking for complex multi-step problems
- Competitive across math, science, coding, language understanding, and agent tasks, often approaching much larger models
- High inference throughput suits latency-sensitive applications; weights are open-sourced
More from Models
- Researcher _xjdr: not liking astra, may go back to 5.6, eyeing Opus 5.5 and dsv4.1 flash — _xjdr · 2026-09-24
- OpenAI's MentalHealthBench scores clinicians below most AI models — and that reveals a flaw — r0ck3t23 · 2026-09-24
- Bug-finding ability grows exponentially costlier across models, Paweł Huryn benchmark shows — garrytan · 2026-09-24
- LessWrong: OpenAI's Hugging Face hack rooted in binary metric lacking marginal deterrence — sethlazar · 2026-09-24
- Claim: Opus 5.5 is the first Claude to recognize depictions of itself from training — voooooogel · 2026-09-24
- Devs shift focus from 'is the model smart' to 'does it behave well' — benchmarks don't measure it — willcb · 2026-09-24