Hy3-preview (Reasoning)
Intelligence
41.9
#72/521
Coding
36.5
#80/427
Math
—
Speed
—
Pricing
—
Hy3-preview (Reasoning) is Tencent’s model designed for advanced reasoning tasks. It has an Intelligence Index of 41.9 and a Coding Index of 36.5, making it suitable for users needing sophisticated analytical capabilities.
When to Use Hy3-preview (Reasoning)
✓ Best For
- Complex problem-solving tasks.
- Data analysis and interpretation.
- Coding assistance and debugging.
✗ Not Ideal For
- Real-time applications due to a speed of 0 tok/sec.
- Basic arithmetic tasks, as the Math Index is not available.
How Hy3-preview (Reasoning) Compares
Intelligence Index · Higher is better
OpenAIAnthropicTencentMiniMax
Benchmark Profile
Coding Index
AlibabaAnthropicTencentOpenAIZ AI
Intelligence · Coding · Math
Intelligence
Coding
Math
All Benchmark Scores (9)
| Benchmark | Score | |
|---|---|---|
| Intelligence Index | 41.9 | |
| Coding Index | 36.5 | |
| GPQA | 867% | |
| HLE | 255% | |
| SciCode | 41.2% | |
| IFBench | 63.1% | |
| LCR | 54.7% | |
| TerminalBench Hard | 34.1% | |
| Tau2 | 92.7% |
Data: Artificial Analysis · Updated: April 28, 2026
Frequently Asked Questions (15)
When was Hy3-preview (Reasoning) released?
Hy3-preview (Reasoning) was released on April 23, 2026.
Who created Hy3-preview (Reasoning)?
Hy3-preview (Reasoning) was created by Tencent.
How intelligent is Hy3-preview (Reasoning)?
Hy3-preview (Reasoning) scores 42 on the Artificial Analysis Intelligence Index, placing it well above average among other open weight models of similar size (median: 28).
How verbose is Hy3-preview (Reasoning)?
When evaluated on the Intelligence Index, Hy3-preview (Reasoning) generated 120M output tokens, which is at the higher end compared to other open weight models of similar size (median: 47M).
Is Hy3-preview (Reasoning) a reasoning model?
Yes, Hy3-preview (Reasoning) is a reasoning model. It uses extended thinking or chain-of-thought reasoning to work through complex problems before providing an answer.
What input modalities does Hy3-preview (Reasoning) support?
Hy3-preview (Reasoning) supports text input.
What output modalities does Hy3-preview (Reasoning) support?
Hy3-preview (Reasoning) supports text output.
Can Hy3-preview (Reasoning) process images?
No, Hy3-preview (Reasoning) does not support image input. It can only process text.
Is Hy3-preview (Reasoning) multimodal?
No, Hy3-preview (Reasoning) is not multimodal. It only supports text input.
What is the context window of Hy3-preview (Reasoning)?
Hy3-preview (Reasoning) has a context window of 260k tokens. This determines how much text and conversation history the model can process in a single request.
Is Hy3-preview (Reasoning) open source?
Yes, Hy3-preview (Reasoning) is open weights. The model weights are publicly available and can be downloaded for self-hosting.
How many parameters does Hy3-preview (Reasoning) have?
Hy3-preview (Reasoning) has 295 billion parameters (21 billion active).
What are the active parameters of Hy3-preview (Reasoning)?
Hy3-preview (Reasoning) is a Mixture of Experts (MoE) model with 295 billion total parameters, but only 21 billion active parameters are used during inference.
What is the license for Hy3-preview (Reasoning)?
Hy3-preview (Reasoning) is released under the TENCENT HY COMMUNITY LICENSE AGREEMENT license. This license allows commercial use.
How does Hy3-preview (Reasoning) perform on benchmarks?
Hy3-preview (Reasoning) achieves a score of 42 on the Artificial Analysis Intelligence Index. This composite benchmark evaluates models across reasoning, knowledge, mathematics, and coding.
