$ diff hy4-preview deepseek-v4-pro-0813
Hy4 preview vs DeepSeek V4-Pro-0813
Values come from Hy4 preview and DeepSeek V4-Pro-0813, where each is cited to its source. This page states no benchmark result and ranks neither model — it puts two published specifications next to each other. Where a lab has not published a figure, the row says so rather than guessing.
TencentHy4 preview
- context
- 1M
- weights
- open
- $/M in
- $0.834
- $/M out
- $2.501
- context
- 1.0M
- weights
- open
- $/M in
- $0.66
- $/M out
- $1.98
What actually differs
- Context window
- DeepSeek V4-Pro-0813 takes 1.0M against 1M — modestly more room in a single request.
- Input price
- DeepSeek V4-Pro-0813 at $0.66/M against $0.834/M — 1.3× cheaper to feed.
- Output price
- DeepSeek V4-Pro-0813 at $1.98/M against $2.501/M — 1.3× cheaper to generate. Output dominates the bill on most agentic workloads, where the model writes far more than it reads.
- Weights
- Both publish weights — Hy4 preview under Apache-2.0, DeepSeek V4-Pro-0813 under MIT (open-weight, as reported). Check the licences rather than assuming they permit the same commercial use.
- Recency
- Hy4 preview shipped 15 days after DeepSeek V4-Pro-0813 (2026-08-28 vs 2026-08-13).
Full spec
| Attribute | Hy4 preview | DeepSeek V4-Pro-0813 |
|---|---|---|
| Developer | Tencent | DeepSeek |
| Released | 2026-08-28 | 2026-08-13 |
| Context window | over 1,000,000 | 1,048,576 (1M) — 384K max output |
| Pricing | $0.834/M input · $2.501/M output | $0.66/M input (cache miss) · $1.98/M output off-peak; $1.32 · $3.96 at peak — see Pricing |
| License | Apache-2.0 | MIT (open-weight, as reported) |
| Availability | Hugging Face (BF16 + FP8), ModelScope, GitCode, CNB, Tencent Cloud TokenHub, OpenRouter, WorkBuddy, CodeBuddy, Yuanbao, ima | DeepSeek API (Responses format, Codex-adapted), Hugging Face (`deepseek-ai/DeepSeek-V4-Pro-0813`) |