The 17 October 2026 build of DeepSeek V4.1 Flash, refusals removed, under its own id so it never changes under you. Private, OpenAI-compatible, reads pictures, and answers what it’s asked.
Easy? You came to my city wearing my colours, and you want easy? Sit down. You’re going to lose slowly.Easy? You came to my city wearing my colours, and you want easy? Sit down. You’re going to lose slowly.The man on the left. He’s smiling with his mouth, not his eyes, and his hand is already on the door.There’s nothing above your message. No system prompt, no instructions. Just you.Easy? You came to my city wearing my colours, and you want easy? Sit down. You’re going to lose slowly.
streaming…deepseek-v4.1-flash-1017-uncensored · 1 credit · prompt and reply not keptdeepseek-v4.1-flash-1017-uncensored · 1 credit · prompt and reply not keptdeepseek-v4.1-flash-1017-uncensored · 1 credit · prompt and reply not keptdeepseek-v4.1-flash-1017-uncensored · 1 credit · prompt and reply not kept
Words and pictures
DeepSeek V4.1 Flash 1017 Uncensored
DeepSeek’s 17 October build under its own id: pictures, tools and thinking, and the same build every call.
Model
deepseek-v4.1-flash-1017-uncensored
Reads
Text, pictures
Context
1M tokens
Longest reply
128K tokens
Tools, thinking
Yes · none, low, high, max
Per 1M tokens
$0.45 in · $0.011 cached · $1.75 out
Why use the dated id
Replies from the same build while you tune your prompts
Evals you can rerun against the same weights
Apps that ship one fixed model and move when they choose
Everything the undated id does: pictures, tools, a million tokens
Nothing added to your prompt.
Nothing cut from the reply.
Nothing kept after.
Paid with nokeep credits you buy: 1,000 are $1, from $5. Billed by the token, to a millionth of a credit, with the fraction carried to your next call. Refused calls, and calls that fail before a word, cost nothing; a reply cut off mid-stream is billed for what it wrote. A plan’s credits stay in the app.
Prices and limits checked 26 October 2026
Call the 1017 build
deepseek-1017
stream = client.chat.completions.create(
model="deepseek-v4.1-flash-1017-uncensored",
messages=[{"role": "user", "content": "Play the rival."}],
stream=True, # answers at once; reasoning_effort="high" to have it think
)
for chunk in stream:
if chunk.choices:
print(chunk.choices[0].delta.content or "", end="")
DeepSeek V4.1 Flash 1017 Uncensored2,700+ replies$0.0018 each
A reply here is 3,000 tokens in and 250 out, thinking off. Longer chats cost more per reply, since the history is sent each time.
Questions, answered.
How is it different from deepseek-v4.1-flash-uncensored?+
Today they run the same build. The undated id moves to each new build of DeepSeek V4.1 Flash Uncensored as we switch to it; this id stays on the 17 October 2026 build.
Is this the same model as DeepSeek V4.1 Flash?+
The same weights as its 17 October 2026 build, with the refusal direction removed. It answers like the original on everyday work and stops declining what it’s asked.
What if the 1017 build is retired?+
The [changelog](/docs/changelog) will say so, with the date and the build to move to.
What does it cost?+
$0.45 per million tokens in and $1.75 out, paid with nokeep credits (1,000 credits are $1). The start of a prompt the model has just read counts as cached, at $0.011 per million, so a long chat or an agent that resends its history pays far less. Billed to a millionth of a credit, with the fraction carried to your next call.
Can it read pictures?+
Yes, up to ten per request, as image_url content parts: a link or a data: URL, in the OpenAI format. It doesn’t read video.
Is it private?+
We add nothing to your prompt, cut nothing from the reply and keep only token counts for the bill. The model runs on a private tier that stores no prompts or replies.