DeepSeek-V4-Flash-Base

Name: DeepSeek V4 Flash Base
Author: DeepSeek-AI

Foundation modelDeepSeek V4

DeepSeek V4 Flash Base

Release date: 2026-04-24Knowledge cutoff: 2025-0521

Live demoGitHubHugging Face Compare

Parameters

284B

Context length

Chinese support

Supported

Reasoning ability

DeepSeek 于 2026 年 4 月 24 日发布的 DeepSeek-V4 Flash Base checkpoint，284B 总参数、13B 激活参数、1M 上下文，MIT 许可。

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

DeepSeek-V4-Flash-Base

Model basics

Reasoning traces

Not supported

Thinking modes

Thinking modes not supported

Context length

1M tokens

Max output length

375K tokens

Model type

Foundation model

Modality (in / out)

Text → Text

Release date

2026-04-24

Model file size

No data

MoE architecture

Yes

Total params / Active params

284B / 13B

Knowledge cutoff

2025-05

DeepSeek-V4-Flash-Base

Open source & experience

Code license

MIT License

Weights license

MIT License- 免费商用授权

GitHub repo

GitHub link unavailable

Hugging Face

https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-Base

Live demo

No live demo

DeepSeek-V4-Flash-Base

Official resources

Paper

DeepSeek-V4 Technical Report

DataLearnerAI blog

No blog post yet

DeepSeek-V4-Flash-Base

API details

API speed

5/5

No public API pricing yet.

DeepSeek-V4-Flash-Base

Benchmark Results

DeepSeek-V4-Flash-Base currently shows benchmark results led by LongBench v2 (11 / 11, score 44.70). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.