Submit
TencentOpen-weight·1000K context·770B params·Apache-2.0

Tencent Hy4 Previewtencent/hy4-preview

Tencent Hy4 Preview is an open-weight MoE model for coding, long-context reasoning and document workflows. The model card reports 770B total parameters and 49B active per token, plus a native multi-token-prediction layer. This is an early preview with documented limitations.

Availability and billing notes

Reviewed September 21, 2026 · check the linked provider documentation for later changes.

Weights are available for self-hosting. Tencent also announces hosted access through Tencent Cloud TokenHub and OpenRouter.

The launch announcement quotes $0.834 input, $0.042 cached input and $2.501 output per million tokens. A current provider-specific endpoint and billable model ID were not independently matched in this review, so these launch figures are not entered as a verified hosting offer.

Official source 1 →Official source 2 →

Pricing across providers

No current verified API pricing is listed. See the official model documentation for availability and billing.

Capabilities

  • code
  • reasoning
  • long_context

Technical specs

Context
1000K
Max output
—
Parameters
770B
Release
2026-08-28
Training cutoff
—
License
Apache-2.0
Hugging Face →Official documentation →

Similar models

Compare with

  • Tencent Hy4 Preview vs DeepSeek V3.1
    Comparison planned — not yet published
  • Tencent Hy4 Preview vs R1
    Comparison planned — not yet published
  • Tencent Hy4 Preview vs DeepSeek V3
    Comparison planned — not yet published

Frequently asked

How much does Tencent Hy4 Preview cost?+
No current verified API pricing is listed. This does not mean the model is free or that a hosted API is currently available. Check the official model documentation below.
Is Tencent Hy4 Preview open-source? Can I fine-tune it?+
Tencent Hy4 Preview has published weights. License: Apache-2.0. Review the linked repository and license for deployment, fine-tuning and commercial-use conditions.
What's the maximum context window for Tencent Hy4 Preview?+
The model supports up to 1,000,000 tokens of context (input + output). Some hosted versions may impose a smaller limit — check the "Context" column in the pricing matrix for each provider.