Hy4 preview is a mixture-of-experts model from Tencent with 49B active parameters and a 1M-token context window, designed for coding, tool use, and multi-step reasoning tasks.
- Strengths
- Handles long contexts and complex multi-tool workflows effectively, with strong performance on coding and agentic tasks requiring extended reasoning chains.
- Best for
- Building coding agents, orchestrating complex tool-use pipelines, and productivity applications that need to reason over large documents or conversation histories.
- Limitations
- As a preview release, stability and performance characteristics may change; the Hy3 preview (released earlier in 2026) and Hy3 (released 2026-07) offer alternative reasoning-focused options with different architectural tradeoffs.
Input / 1M
$0.834
Output / 1M
$2.50
Cached input / 1M
$0.042
Context window
1.05M
Price history
Snapshots
| Effective | Input | Output | Cached in | Note | Source |
|---|---|---|---|---|---|
| 28 Aug 2026 | $0.834 | $2.50 | $0.042 | Imported from OpenRouter | openrouter.ai |