Tencent Hy4 Preview and Qwen 3.8 Flash on API

Two models from Chinese labs were added to GetVibe API on September 24, 2026: Tencent Hy4 Preview and Qwen 3.8 Flash from Alibaba's Qwen team. Both are text models with reasoning and tool calling, and both work through the same OpenAI-compatible endpoint as the rest of the catalog.

Tencent Hy4 Preview

Hy4 Preview is Tencent's preview text model for reasoning and agentic tasks. As a preview release it can change over time, so pin it in tests and watch the model page for updates.

• Model ID: hy4-preview
• Context window: 262,144 tokens
• Reasoning: two levels, low and high, billed as output tokens
• Tool calling and JSON mode: yes
• Price: $0.0334 input · $0.1001 output per 1M tokens

Qwen 3.8 Flash

Qwen 3.8 Flash is the fast member of the Qwen 3.8 family. It is a mixture-of-experts model built for high-volume text work such as tagging, extraction, summaries and first-pass drafts.

• Model ID: qwen3.8-flash
• Context window / max output: 256K / 32K tokens
• Reasoning: thinking mode, billed as output tokens
• Tool calling and JSON mode: yes
• Knowledge cutoff: June 2025
• Price: $0.1039 input · $0.3254 output per 1M tokens

If you need more depth from the same family, Qwen3.8 Max (qwen/qwen3.8-max, 256K context) is also in the catalog at $0.1543 input · $0.4629 output per 1M tokens.

Try them

curl https://getvibeapi.com/v1/chat/completions \
  -H "Authorization: Bearer $GETVIBE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.8-flash",
    "messages": [{"role": "user", "content": "Extract the company names from this text: ..."}]
  }'

Replace qwen3.8-flash with hy4-preview to call Tencent's model. Nothing else changes. For agent loops, see tool calling. For strict machine-readable output, see structured outputs.

Which one when

• Qwen 3.8 Flash: large batches of short, well-defined text jobs where price per request drives the choice.
• Tencent Hy4 Preview: tasks that benefit from explicit reasoning, where the high level lets you trade tokens for depth.

Test both on a sample of your own prompts before you switch production traffic. The article how to evaluate an AI API provider has a short checklist. Live latency and cached-input prices are on the Hy4 Preview and Qwen 3.8 Flash model pages.