Hy4 Preview API
- Input
- $0.834per 1M tokens
- Output
- $2.501per 1M tokens
- Cached input
- $0.042per 1M tokens
- Context
- 1M
Hy4 Preview for OpenAI-style Chat Completions, Anthropic-style Messages, OpenAI-style Responses.
Explore Hy4 Preview for OpenAI-style Chat Completions, Anthropic-style Messages, OpenAI-style Responses through the unified API. Supported interfaces and current availability are shown below.
- OpenAI-style Chat Completions
- Anthropic-style Messages
- OpenAI-style Responses
Model details
- Context window
- 1M tokens
- Max output
- 66K tokens
- Input
- Text
- Output
- Text
- Capabilities
- Reasoning, Tool calling, Streaming, JSON output
- Interfaces
- OpenAI-style Chat Completions, Anthropic-style Messages, OpenAI-style Responses
Hy4 Preview API pricing
Per 1M tokens. The Inferray price is what you are charged — no gateway fee is added on top of it.
| Metric | Price |
|---|---|
| Input | $0.834 |
| Cached input | $0.042 |
| Output | $2.501 |
What Hy4 Preview costs for real work
Typical workloads at list price, worked out from the rates above. Nothing is added on top.
| Workload | Tokens | Cost |
|---|---|---|
| 1,000 chat replies | 1,500 tokens in, 400 out, each | $2.25 |
| Summarise a 100-page document | 60K tokens in, 2K out | $0.055 |
| 10M tokens of coding-agent work | 10M tokens: 85% cached input, 12% fresh input, 3% output | $2.11 |
| 1M output tokens | Generated text only | $2.50 |
Hy4 Preview is pay-as-you-go only
Hy4 Preview is not included in Token Plans. Call it from a pay-as-you-go balance at the list price above, with no gateway fee.
Why teams pick it
- Hy4 Preview for OpenAI-style Chat Completions, Anthropic-style Messages, OpenAI-style Responses.
Call Hy4 Preview
Hy4 Preview answers on any of the APIs below — pick the one your code already speaks and just swap in your Inferray key.
curl https://api.inferray.com/v1/chat/completions \
-H "Authorization: Bearer inferray_..." \
-H "Content-Type: application/json" \
-d '{
"model": "hy4-preview",
"messages": [
{
"role": "user",
"content": "Hello"
}
]
}'curl https://api.inferray.com/v1/messages \
-H "Authorization: Bearer inferray_..." \
-H "Content-Type: application/json" \
-d '{
"model": "hy4-preview",
"max_tokens": 1024,
"messages": [
{
"role": "user",
"content": "Hello"
}
]
}'curl https://api.inferray.com/v1/responses \
-H "Authorization: Bearer inferray_..." \
-H "Content-Type: application/json" \
-d '{
"model": "hy4-preview",
"input": "Hello"
}'Conversation compaction is also available at POST /v1/responses/compact.
Use Hy4 Preview in your tools
Every coding agent below has a setup guide. Set the model to hy4-preview.
More from Tencent Cloud
Alternatives to Hy4 Preview
Questions about this model
- How much does the Hy4 Preview API cost?
- Hy4 Preview costs $0.834 per 1M input tokens and $2.501 per 1M output tokens on Inferray, and cached input is $0.042 per 1M. That is the list price, with no gateway fee on top.
- Is there a free tier for Hy4 Preview?
- No. You add credit and pay Hy4 Preview's list rate for each request — seats, projects, and API keys are free, and pay-as-you-go credit never expires.
- Am I charged for failed requests?
- No. A request that fails without returning a result is not billed. If a response fails part-way, the tokens already processed before the failure are charged.
- Is Hy4 Preview included in Token Plans?
- No. Hy4 Preview is pay-as-you-go only; Token Plans cover the other text models in the catalog.
- Can I use Hy4 Preview in Claude Code?
- Yes. Set ANTHROPIC_BASE_URL to https://api.inferray.com, ANTHROPIC_AUTH_TOKEN to your Inferray key, and ANTHROPIC_MODEL to hy4-preview. The same key works in Codex, opencode, Cline, and other agents.
- Does Hy4 Preview support tool calling and vision?
- Hy4 Preview supports tool calling, reasoning, JSON output and streaming on Inferray. It does not take image input.
- Is it the real Hy4 Preview?
- Yes. A request for hy4-preview is answered by Hy4 Preview from Tencent Cloud, billed at its published list price. Inferray never substitutes a different or smaller model.
Prices from the catalog as of November 12, 2026.