Skip to main content
DeepSeekFlash, open weights

DeepSeek V4.1

DeepSeek's fast, open-weight model with vision, for low-cost everyday work.

Made on Runable

Each clip is one prompt, run as a single Runable task.

DeepSeek 4.1 on Runable

  • Fast builds

    About 206 tokens per second through pages, scripts and drafts.

  • Reads images

    Hand it screenshots, product photos or charts.

  • Mix with premium

    Draft on DeepSeek, finish on Opus 5.5.

How to use DeepSeek 4.1 on Runable

  1. Select DeepSeek 4.1

    Start a task on Runable and pick DeepSeek 4.1 in the model menu.

  2. Describe the job

    Type what you need, or add files and links. Runable plans it and gets to work.

  3. Edit and ship

    Refine by chatting, then publish the site or export to PPTX, PDF or Google Slides.

One prompt, finished work

Real results from single Runable tasks. Pick one to watch it.

Made on Runable
Build an award-winning single-page site for an environmental nonprofit called Verdant: an animated 3D Earth hero, live stats, and scroll-triggered animations.
token context window
1M
max output tokens
384K
tokens per second (Artificial Analysis)
206
turn to build
Your

Why choose DeepSeek 4.1

  • A fraction of the cost

    $0.30 / $1.20 per 1M tokens at peak, half off-peak.

  • Beats its own flagship

    Ahead of DeepSeek’s V4-Pro, says DeepSeek.

  • Strong on agent tasks

    77.0 on DeepSWE v1.1 and 60.5 on AutomationBench.

  • Very long answers

    Up to 384K output tokens per reply.

  • Open weights

    MIT licence, on Hugging Face.

  • Best in its class

    #2 of 121 in its Artificial Analysis class.

DeepSeek 4.1 specs

Maker
DeepSeek
Model
DeepSeek-V4.1-Flash (updated as V4.1-Flash-1017 on Oct 17, 2026)
Released
September 10, 2026
API model ID
deepseek-flash
Context window
1M tokens
Max output
384K tokens
Input / output
Text and images in, text out
Reasoning
Thinking and non-thinking modes; effort low, high or max
Architecture
Mixture-of-experts, open weights (MIT licence)
API price
$0.30 input / $1.20 output per 1M tokens (peak); half off-peak
Runable’s picker shows this model as “DeepSeek 4.1”. There is no model officially called DeepSeek 4.1; DeepSeek’s released V4.1 model is V4.1 Flash. API prices are DeepSeek’s list prices; on Runable you use your plan’s credits instead.

The only AI agent you need

Compare plansHelp

Need SSO, data localization, premium support, onboarding assistance? Contact Sales

DeepSeek 4.1 FAQs

DeepSeek V4.1 Flash is DeepSeek’s newest model, released on September 10, 2026 and updated on October 17, 2026. DeepSeek calls it the smallest model in its new architecture family, with native visual understanding, and says it beats flagship models including its own V4-Pro.

Sources: DeepSeek: V4.1 Flash release (Sep 10, 2026) · DeepSeek: V4.1-Flash-1017 update (Oct 17, 2026) · DeepSeek API pricing and models · Hugging Face: DeepSeek-V4.1-Flash-1017 · Artificial Analysis: DeepSeek V4.1 Flash 1017

More AI models on Runable

Put DeepSeek 4.1 to work

Describe the job. Runable runs it on DeepSeek 4.1 and ships the result.