Model Library
TencentReleased April 23, 2026Hy3-preview
Tencent · Released April 23, 2026
Intelligence score
34
At a glance
- Intelligence
- 34
- Speed
- 196.2tokens/sec
- Latency
- 3.15stime to first token
- Input price
- $0.12per 1M tokens
- Output price
- $0.43per 1M tokens
- Context window
- 260Ktokens
- Parameters
- 295billions
Summary
Comparison summary
Hy3-preview (Reasoning) scores 34 (estimated) on the Artificial Analysis Intelligence Index, placing it well above average among other open weight models of similar size (median: 25). Hy3-preview (Reasoning) generates output at 196.2 tokens per second (based on the median across providers serving the model), which is well above average compared to other open weight models of similar size (median: 61.7 t/s). Hy3-preview (Reasoning) costs $0.12 per 1M input tokens (very competitive, median: $0.59) and $0.43 per 1M output tokens (very competitive, median: $2.20), based on the median across providers serving the model.
Specifications
Technical specifications
| Model name | Hy3-preview |
|---|---|
| Provider | Tencent |
| Release date | April 23, 2026 |
| Intelligence score | 34 |
| Speed | 196.2 tokens/sec |
| Latency (TTFT) | 3.15s |
| Input price | $0.12 / 1M tokens |
| Output price | $0.43 / 1M tokens |
| Context window | 260,000 tokens |
| Parameters | 295B |
| Reasoning | Yes |
| Input modalities | text |
| Output modalities | — |
| Open weights | Yes |
| License | TENCENT HY COMMUNITY LICENSE AGREEMENT |
| API availability | — |
| Source | Model Intelligence legacy profile: hy3 |
Pricing
Token prices
Price per 1M tokens
Hy3-preview, as listed in the catalog record
- Input$0.12
- Output$0.43
| Token type | Price per 1M |
|---|---|
| Input | $0.12 |
| Output | $0.43 |
Model Intelligence legacy profile: hy3
FAQ · 20
Frequently asked questions
Hy3-preview (Reasoning) was released on April 23, 2026.
Hy3-preview (Reasoning) was created by Tencent.
Hy3-preview (Reasoning) scores 34 (estimated) on the Artificial Analysis Intelligence Index, placing it well above average among other open weight models of similar size (median: 25).
Hy3-preview (Reasoning) generates output at 196.2 tokens per second (based on the median across providers serving the model), which is well above average compared to other open weight models of similar size (median: 61.7 t/s).
Hy3-preview (Reasoning) has a time to first token (TTFT) of 3.15s (based on the median across providers serving the model), which is at the higher end compared to other open weight models of similar size (median: 2.01s).
Hy3-preview (Reasoning) costs $0.12 per 1M input tokens (very competitive, median: $0.59) and $0.43 per 1M output tokens (very competitive, median: $2.20), based on the median across providers serving the model.
Hy3-preview (Reasoning) costs $0.12 per 1M input tokens and $0.43 per 1M output tokens (based on the median across providers serving the model). For a blended rate (7:2:1 cache hit/input/output ratio), this is $0.10 per 1M tokens. Pricing may vary by provider.
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.
Hy3-preview (Reasoning) supports text input.
Hy3-preview (Reasoning) supports text output.
No, Hy3-preview (Reasoning) does not support image input. It can only process text.
No, Hy3-preview (Reasoning) is not multimodal. It only supports text input.
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.
Yes, Hy3-preview (Reasoning) is open weights. The model weights are publicly available and can be downloaded for self-hosting.
Hy3-preview (Reasoning) has 295 billion parameters (21 billion active).
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.
Hy3-preview (Reasoning) is released under the TENCENT HY COMMUNITY LICENSE AGREEMENT license. This license allows commercial use.
Hy3-preview (Reasoning) achieves a score of 34 on the Artificial Analysis Intelligence Index. This composite benchmark evaluates models across reasoning, knowledge, mathematics, and coding.
Yes, Hy3-preview (Reasoning) is available via API through 2 providers.
Hy3-preview (Reasoning) is available through 2 API providers.