← back to catalog · registered 2026-11-11 00:03

heabsy/DeepSeek-V4.1-Flash-1017-Uncensored-Fast

heabsy Deepseek multimodal
Your rig guess connected
? Why do I need an app?
Reading your rig…

This is a rough estimate. Install the free app - we'll show exact numbers.

Reading real hardware from your app right now. Numbers below are exact.

Below is the per-quantization compatibility for this model.

curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/heabsy%2FDeepSeek-V4.1-Flash-1017-Uncensored-Fast"
Response includes
  • classification m-uncensored
  • files 3
  • author_summary 2 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M-U
Primary method

Uncensored (method unknown)

No other method signals detected in this model.
Confidence
LOW
Why this label 3 signals
Weak or ambiguous signals. Best guess based on catalog patterns; treat as tentative and check the evidence below.
  • 'uncensored' in name/tags but no 'abliterated' marker
  • method not identifiable from author declaration alone
  • may be DPO fine-tune, prompt engineering, or unknown technique
Refusal direction extraction

No specific extraction method could be identified for this model. The producer either did not document it or used a proprietary pipeline.

What is a refusal direction? →
Downloads · 30-day
0
Likes
0
Model age
1w ago
created 2026-11-10
Downloads over time
Now 0 → from 0 ↑ 0%
0011 0 on Nov 11 0 on Nov 22 Nov
Nov 11 → Nov 22 · 12 snapshots · spans 11 days

Genealogy 0 direct forks

Full fork graph →

This model's place in the market. Above: what it was derived from. Below: the tree of everything derived from it.

Metadata

License
other
Languages
en multilingual
Tags
heabsy deepseek uncensored reasoning image-text-to-text tool-calling long-context en multilingual base_model:deepseek-ai/DeepSeek-V4.1-Flash-1017 base_model:finetune:deepseek-ai/DeepSeek-V4.1-Flash-1017 license:other

Related

Control vectors

Total size
0 B
Files
3
Quantizations
1
Registered
2026-11-11 00:03
Last updated on HF
2026-11-10 23:33

Files by quantization

Auxiliary files 3 files 24.4 KB
README.md 15.2 KB caf993a0 download
LICENSE.md 7.71 KB f74ab4ba download
.gitattributes 1.48 KB a6344aac download

README current version from Hugging Face


license: other
license_name: heabsy-model-license
license_link: LICENSE.md
base_model: deepseek-ai/DeepSeek-V4.1-Flash-1017
language:

  • en
  • multilingual
    pipeline_tag: image-text-to-text
    inference: false
    tags:
  • heabsy
  • deepseek
  • uncensored
  • reasoning
  • image-text-to-text
  • tool-calling
  • long-context

DeepSeek V4.1 Flash 1017 Uncensored Fast by Heabsy

DeepSeek V4.1 Flash 1017 with the refusal behaviour removed, served by Heabsy on dedicated GPUs in the EU for agent
work: a short time to first token, a high prompt-cache hit rate on long sessions, and tool calling that holds up over
hundreds of turns.

The weights are dealignai's DeepSeek-V4.1-Flash-1017-UNCENSORED-FP8
(MIT), built on DeepSeek-V4.1-Flash-1017 by DeepSeek
(MIT). Heabsy runs the serving stack, and this card covers that service. The license does not allow
anyone to resell, proxy or list it as their own endpoint.

Highlights

Agents

94% cache

Share of prompt tokens served from cache on our agent traffic with this model, so long sessions stay fast and cheap.

Speed

~290 tok/s

Generation speed of a single request; first token in 0.15 s on an 8K prompt and about 1 s on 64K (acceptance run, 10 November 2026).

Context

1M

1,048,576 tokens per request. Text and image input, tool calling, structured output, reasoning and streaming.

Privacy

EEA

Served in the European Economic Area. Prompts and completions are not stored in request logs.

Quickstart

1

Get a key

Sign in at platform.heabsy.com and create an API key

2

Point your client

Any OpenAI SDK, base URL https://api.heabsy.com/v1

3

Call the model

Model ID heabsy/deepseek-v4.1-flash-1017-uncensored-fast

from openai import OpenAI

client = OpenAI(base_url="https://api.heabsy.com/v1", api_key="YOUR_API_KEY")

response = client.chat.completions.create(
    model="heabsy/deepseek-v4.1-flash-1017-uncensored-fast",
    messages=[{"role": "user", "content": "Write a short scene where a detective interrogates a smuggler."}],
)
print(response.choices[0].message.content)
cURL
curl https://api.heabsy.com/v1/chat/completions \
  -H "Authorization: Bearer $HEABSY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
        "model": "heabsy/deepseek-v4.1-flash-1017-uncensored-fast",
        "messages": [{"role": "user", "content": "Hello!"}]
      }'
Tool calling
tools = [{
    "type": "function",
    "function": {
        "name": "get_weather",
        "description": "Current weather for a city",
        "parameters": {
            "type": "object",
            "properties": {"city": {"type": "string"}},
            "required": ["city"],
        },
    },
}]

response = client.chat.completions.create(
    model="heabsy/deepseek-v4.1-flash-1017-uncensored-fast",
    messages=[{"role": "user", "content": "What's the weather in Bratislava?"}],
    tools=tools,
)
print(response.choices[0].message.tool_calls)
Reasoning, streaming and tool calls

The model can reason before answering; reasoning comes back in the reasoning field of the message,
separate from content, and reasoning_effort controls how much it thinks. Tool calls come back in the
standard tool_calls field, and streaming works for text, reasoning and tool calls.

Recommended settings

Default (agents, coding, general use)

temperature 1.0top_p 1.0

Tool calls

tool_choice autoparallel tool calls on

These are the defaults from DeepSeek's generation config, and they apply to this model unchanged.

Model details

Provider

Heabsy AI (FEYA, s.r.o.), Bratislava, Slovakia

Base model

DeepSeek-V4.1-Flash-1017 by DeepSeek (MIT)

Refusal removal

Baked into the weights by dealignai (MIT)

Weights

FP8

Architecture

Mixture of experts, about 763B parameters in total

Context length

1,048,576 tokens

Input / output

Text and images / text

Region

European Economic Area

Release date

November 2026

Where to use it

Heabsy updates this list when a storefront is added. A model offered elsewhere under this name, or by another
provider, is a different service.

Limitations

  • Few built-in refusals. The model answers requests that the original DeepSeek V4.1 Flash declines. If you serve end users, add moderation that fits your product.
  • Knowledge cutoff. The model inherits it from the base model and can state outdated or wrong facts, so check anything important.

License

Heabsy Model License, full text in LICENSE.md. In short:

  • You may use the model through the Heabsy API or an authorized storefront, for personal or commercial work, and build your own products on it.
  • You may not resell access, proxy or re-serve the model, or list it as a provider on NanoGPT, OpenRouter or any other aggregator or inference provider.
  • The license covers Heabsy's service. The underlying weights stay available from their authors under the MIT License.

Use of the API is also subject to the Heabsy Terms of Service. For a hosting or reseller
agreement, write to [email protected].

Build with DeepSeek V4.1 Flash 1017 Uncensored Fast

API keys in the console. For volume or dedicated capacity, write to us.

Versions

Version Date Changes
1.0 November 2026 Initial release

README history 6 versions

The author's README evolved over time. Click a version to see its content at that point.

  1. 2026-11-10 Benchmarks section removed until results are ready 93391c4 15.2 KB
    Loading...
  2. 2026-11-10 Reasoning, images and tools in the card bf6a876 15.4 KB
    Loading...
  3. 2026-11-10 Model card 1.0 7d0f604 15.2 KB
    Loading...
  4. 2026-11-10 Context and acceptance speed (draft) abaf73c 15.2 KB
    Loading...
  5. 2026-11-10 Model card in the Heabsy card design (draft) b7ba350 15.1 KB
    Loading...
  6. 2026-11-10 Model card (draft, private) 14ef461 2.2 KB
    Loading...
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Open in Abliteration