Tencent: Hy3

TENCENT Developer Architecture Profile

OpenRouter ↗HuggingFace ↗

Intelligence (ELO)1438Chatbot Arena Verified

Max Context Limit262,144Max Out: 128,000 tokens

Prompt Caching75% OFF$0.03 / 1M cached

Standard API / 1M$0.66🧠 Deep Reasoner

Model Capabilities & Contract SLAs

Reasoning
Coding & Logic
Fictional
🧠 Extended Test-Time Reasoning

Hy3 is a 295B-parameter Mixture-of-Experts model from Tencent (21B active, 192 experts with top-8 routing) built for reasoning, agentic workflows, and real-world production use. It supports a configurable reasoning effort:...

Granular Pricing Matrix

Input Tokens (Standard Prompt)$0.13 / 1M

⚡ Cached Input (Prompt Cache Read)75% OFF$0.03 / 1M

Output Tokens (Completion)$0.53 / 1M

Pricing data via OpenRouter. Sync: 10/2/2026

Evaluate Competitors

VS Engine MatchupTencent: Hy3 vs inclusionAI: Ling 3.0 Flash Sante (free)VS Engine MatchupTencent: Hy3 vs ByteDance Seed: Seed-2.0-Code VS Engine MatchupTencent: Hy3 vs Upstage: Solar Pro 4 VS Engine MatchupTencent: Hy3 vs Qwen: Qwen3.7 Flash VS Engine MatchupTencent: Hy3 vs Anthropic: Claude Sonnet 5 VS Engine MatchupTencent: Hy3 vs Qwen: Qwen3.6 Flash