Hy4
Hy4 is a chat-oriented large language model from Tencent, listed in our catalog with limited published specifications.
API Pricing
Cheapest input on OpenRouter · cheapest output on OpenRouter| Provider | Input / 1M | Output / 1M | Cached / 1M |
|---|---|---|---|
| $0.417 | $1.25 | $0.021 | |
| $0.417 | $1.25 | $0.021 | |
| $0.417 | $1.25 | $0.021 | |
| $0.834 | $2.50 | $0.417 | |
| $0.834 | $2.50 | $0.042 | |
| $0.834 | $2.50 | $0.042 |
Prices updated daily. Last check: Nov 5, 2026
Hy4 pricing by provider
Input, output, and batch rates, plus alternatives, for one provider at a time.
Performance & Benchmarks
Source: Artificial Analysis →Benchmarks measured Nov 2026. Scores are independent evaluations, not vendor-reported.
Model Details
General
- Creator
- Tencent
- Modalities
- Text
Capabilities
Strengths & Limitations
Strengths
- Developed by Tencent, a vendor with an established line of large language models and in-house serving infrastructure
- Catalogued as a chat model, so it targets instruction-following and multi-turn dialogue rather than a narrow single-task role
- Listed alongside competing chat models on this page, allowing direct provider-by-provider cost comparison
- Chinese-developer origin may align with workloads that need strong Chinese-language handling, subject to your own evaluation
- Appears in our tracked catalog, so pricing changes across hosting providers are reflected in the table above
Limitations
- Published specifications are thin — we do not have a verified context window for this model
- No benchmark scores are tracked in our database, so capability claims cannot be verified here
- Modality support (for example image input) is not confirmed in our metadata
- Tool calling and structured-output support are not tracked for this entry
- Provider availability may be narrower than for widely hosted Western chat models, which can limit routing and failover options
Key Features
About Hy4
Common Use Cases
As a chat-type model, Hy4 is aimed at assistant-style workloads: answering questions, drafting and rewriting text, summarizing documents, and handling multi-turn conversations in a product chat interface. Because our verified metadata does not include a context window, benchmark results, or confirmed multimodal support, it is best approached as a candidate to benchmark against your existing chat model on representative prompts — particularly if your workload includes Chinese-language content, given Tencent's focus area. For anything requiring a documented long-context limit, confirmed vision input, or guaranteed structured tool-call output, verify those capabilities with the serving provider before committing, since we do not track them for this entry.
Frequently Asked Questions
How much does Hy4 cost to use?
Pricing depends on which provider serves the model and on the pricing type (for example per-input-token versus per-output-token rates, or hourly GPU hosting if you self-host). Rates change frequently, so check the pricing table on this page for current figures across tracked providers.
What is Hy4 best used for?
It is catalogued as a chat model, so the natural fit is assistant-style work: question answering, drafting and editing text, summarization, and multi-turn conversational interfaces. Run it against your own prompt set before adopting it, since published evaluation data is not part of our verified metadata.
Who created Hy4?
Hy4 is from Tencent, the Chinese technology company that also develops the Hunyuan family of AI models.
What context window does Hy4 support?
We do not have a verified context window for Hy4 in our database. Check Tencent's model documentation or your chosen inference provider's API reference, since the maximum context a provider accepts can differ from the model's nominal limit.
Does Hy4 support image input or tool calling?
Neither modality support nor tool-calling support is tracked for this entry in our database, so we cannot confirm either way. Consult the provider's API documentation before building features that depend on vision input or structured function calls.