DeepSeek V4 Pro 0813 上线 OpenRouter，仅限 API 调用

DeepSeek V4 Pro 0813 (on OpenRouter)

AI 导读

DeepSeek 最新 Pro 模型 V4 Pro 0813 已上线，仅通过 API 提供，目前可在 OpenRouter 上访问。该模型在低、中、高三种推理强度下生成了差异明显的鹈鹕图像，作者称在其他模型中未见此类现象。官方尚未确认是否开源权重，但鉴于 4 月和 7 月版本均已发布权重，开源可能性较大。

正文 · 原文

The latest DeepSeek Pro model is now available, via API only. I had to link to OpenRouter because DeepSeek don't have any obvious announcement page for their new model.

I haven't been able to confirm if they plan to release the open weights, but given the weights are available for both April's deepseek-ai/DeepSeek-V4-Pro and July's deepseek-ai/DeepSeek-V4-Flash-0731 it seems likely.

Interestingly I got very different looking pelicans for the three different reasoning levels of low, medium, and high. I've not noticed this kind of difference from any other model:

Low:

Flat vector illustration of a white pelican with a large orange beak, wearing a straw hat with an orange band, riding a teal road bicycle in profile, set against a pale cream circle with a dashed outline and small motion marks trailing behind.

Medium:

A similar cartoon pelican cycling, drawn in a looser outlined style: the bird's body is mostly white line art, its orange beak pouch hangs open under a yellow cap, a long red tongue streams backwards towards a yellow sun, and a small blue fish sits on a tray by the handlebars of a green bicycle whose wheels are drawn as broken yellow arcs.

High:

The pelican again, this time on a red bicycle against a pale blue background, with a bright yellow beak and pouch, a purple pennant flag on the back, a wicker front basket holding a small fish, and black musical notes floating in the top right corner.

In terms of benchmarks... as far as I can tell those were released to the Official DeepSeek WeChat Group, then copied and pasted into a post on Reddit which was deleted by the moderators for being "low-effort", then copied into this ASCII-art table on Hacker News.