Models Model details

Tencent

Tencent: Hy3

tencent/hy3

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: a direct no-think mode by default, plus low and high chain-of-thought modes for complex math, coding, and multi-step problems. With a 256K context window, Hy3 targets long-horizon tasks, including improved coreference resolution, multi-turn constraint tracking, and stable tool-calling that generalizes across agent scaffoldings. Tencent positions it as a reliable, cost-effective option across coding, document processing, financial analysis, game development, and frontend design, with a strong emphasis on grounded, anti-hallucination behavior that answers when grounded and flags when evidence is missing rather than fabricating.

Input
$0.132

/ 1M tokens

Output
$0.528

/ 1M tokens

Context window
262.14K

tokens

Limits

Context

262,144

tokens

Max output

128,000

49% of the context window

Parameters

298,786,155,776

Model facts

Author

Tencent

Tokenizer

Other

Modality

Text -> Text

License

Apache-2.0

Created

Jul 6, 2026

Hugging Face

tencent/Hy3

Default generation parameters

temperature

0.9

top_p

1

Reasoning

Default HighHighLowNone

Publication status

Hugging Face: verifiedLicense: publishedParameter count: published

Modalities

Input · TextOutput · Text

Generation controls

Frequency penaltyInclude reasoningLogit biasMax completion tokensMax tokensMin pPresence penaltyReasoningReasoning effortRepetition penaltyResponse formatSeedStopStructured outputsTemperatureTool choiceToolsTop kTop p

Creative tasks

Category performance

3D
Top 51%

1176 Elo

44.0% win

Websites
Top 58%

1183 Elo

41.3% win

Code
Top 56%

1178 Elo

41.2% win

UI components
Top 63%

1163 Elo

40.2% win

Games
Top 65%

1139 Elo

38.6% win

Data visualization
Top 77%

1125 Elo

36.1% win