DeepSeek: DeepSeek V4.1 Flash 1017NEW

Name: DeepSeek V4.1 Flash 1017
Brand: DeepSeek
Price: 16.0468 RUB
Availability: InStock

deepseek/deepseek-v4.1-flash-1017

textVisionReasoning

DeepSeek V4.1 Flash 1017 is the October 17, 2026 snapshot of DeepSeek V4.1 Flash, served by DeepSeek under the `deepseek-flash` API model name. V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture. It activates 8B parameters on input and 16B on output from a 552B-parameter backbone, an asymmetric split that keeps per-token compute low relative to the model's total size. Image understanding is native to the architecture, with visual and text embeddings trained jointly from the start of pre-training rather than added afterward as in the earlier experimental V4 Flash Vision Exp. It is suited for coding, terminal, and computer-use agents, along with long-horizon tasks that must run to completion across many steps and long-context analysis. Compressed KV caching cuts cache memory to roughly a quarter of the previous Flash generation, significantly reducing costs on agentic workloads. DeepSeek positions it as the cost-efficient tier of the V4.1 family and reports that it exceeds V4 Pro on performance, speed, and task completion time.

Price in / out

Cache, read / write

Context / Max output

1.05M / 393K

Released

Oct 17, 2026

Additional rates

Flex, in / out

During some UTC hours the upstream charges 2× more — per-route rates are in the table below.

Parameters

messagesrequired

reasoning_effortoff · low · high · max

max_tokensoptional · alias max_completion_tokens

streamtrue · false

toolsfunction calling (OpenAI format)

tool_choiceauto · none · required

response_formatjson_object · json_schema

web_search · · ·

providerrouting: order · only · ignore · sort

service_tierflex · priority

stream_options{"include_usage": true}

image_urlimage (data: URL)

Fine-tuning · 11

temperature0–2

stopoptional

seedoptional

top_p0–1

top_koptional

min_p0–1

frequency_penalty−2…2

presence_penalty−2…2

repetition_penalty0–2

logprobstrue · false

top_logprobs0–20

How to use via API

from openai import OpenAI

client = OpenAI(base_url="https://api.mixen.ai/v1", api_key="mxn-...")
resp = client.chat.completions.create(
    model="deepseek/deepseek-v4.1-flash-1017",
    messages=[{"role": "user", "content": "Привет"}],
    reasoning_effort="low",  # off · low · high · max
    temperature=0.7,
    max_tokens=1024,
    stream=False,
)
print(resp.choices[0].message.content)

# Стриминг: stream=True — ответ приходит чанками
for chunk in client.chat.completions.create(
    model="deepseek/deepseek-v4.1-flash-1017",
    messages=[{"role": "user", "content": "Привет"}],
    stream=True,
):
    print(chunk.choices[0].delta.content or "", end="", flush=True)

# эконом-тир: тот же ответ вдвое дешевле, но ниже приоритет в очереди
cheap = client.chat.completions.create(
    model="deepseek/deepseek-v4.1-flash-1017",
    messages=[{"role": "user", "content": "Привет"}],
    extra_body={"service_tier": "flex"},
)

API docsTry in the bot

Related models