NOUMENOpen app
NOUMEN

DeepSeek · Open weights

DeepSeek R1 Qwen Abliterated

Added Jan 1, 2024Open weights

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Use in the app

Uncensored version of the Deepseek R1 Qwen 32B model

  • Reasoning
Input
Text
Output
Text
Discovery tier
Open weights

Pricing

USD per 1M tokens
Input
$1.40
Output
$1.40

Reference prices as published by the model provider. API example

Specifications

Context window
16.4K
Max output
8.2K
Average output (7 days)
200 tokens
Longer than 21% of 468 models

Side by side

DeepSeek R1 Qwen Abliterated compared

Specifications, prices, speed and benchmark scores next to each other. Benchmarks below show every compared model too.

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

DeepSeek R1 Qwen AbliteratedDeepSeekThis modelDeepSeek V4.1 Flash 1017 ThinkingDeepSeekRemove
ProviderDeepSeekDeepSeek
Discovery tierOpen weightsOpen weights
AddedJan 1, 2024Oct 17, 2026
Limits
Context window16.38K1.05MBest
Max output8.19K393.22KBest
Average output200 tokens1.2K tokens
Pricing, USD per 1M tokens
Input$1.40$0.10Best
Output$1.40$0.60Best
Cache read—$0.003
Performance
Throughput15.6 tok/s206.5 tok/sBest
Time to first token12.4 s2.3 sBest
Request success (24h)58.0%99.8%Best
Benchmark indexes
Intelligence Index—45.1
Capabilities
ReasoningYesYes
ToolsNoYes
VisionNoYes
Structured outputNoYes
Reasoning effort—none, low, high, max

Public evaluations

Benchmarks

Independent evaluation results, with how each score ranks among the models in the Noumen catalog.

  • DeepSeek R1 Qwen Abliterated
  • DeepSeek V4.1 Flash 1017 Thinking

Intelligence Index

  • DeepSeek R1 Qwen Abliterated—
  • DeepSeek V4.1 Flash 1017 Thinking45.1
3.8Catalog models by score55.1

Agentic work

  • AutomationBench-AA

    Workflow automation, penalized for guardrail violations

    • DeepSeek R1 Qwen Abliterated: —
    • DeepSeek V4.1 Flash 1017 Thinking: 66.6%

  • AutomationBench-AA tasks completed

    Workflows finished end to end without a guardrail violation

    • DeepSeek R1 Qwen Abliterated: —
    • DeepSeek V4.1 Flash 1017 Thinking: 36.7%

  • AA-Briefcasev1.1

    Agentic knowledge work, rated head to head

    • DeepSeek R1 Qwen Abliterated: —
    • DeepSeek V4.1 Flash 1017 Thinking: 1,570 Elo

  • GDPval-AAv2.1

    Economically valuable occupational tasks, rated head to head

    • DeepSeek R1 Qwen Abliterated: —
    • DeepSeek V4.1 Flash 1017 Thinking: 1,746 Elo

Document reasoning

  • GDP.pdf

    Professional reasoning over PDF documents, all-pass rate

    • DeepSeek R1 Qwen Abliterated: —
    • DeepSeek V4.1 Flash 1017 Thinking: 17.2%

  • AA-LCRv1.1

    Reasoning across long documents

    • DeepSeek R1 Qwen Abliterated: —
    • DeepSeek V4.1 Flash 1017 Thinking: 82.7%

  • MLCR-AA

    Long-context reasoning over medical records

    • DeepSeek R1 Qwen Abliterated: —
    • DeepSeek V4.1 Flash 1017 Thinking: 26.7%

Reasoning and knowledge

  • Humanity’s Last Exam

    Expert-level questions across academic fields

    • DeepSeek R1 Qwen Abliterated: —
    • DeepSeek V4.1 Flash 1017 Thinking: 43.3%

Coding

  • Terminal-Benchv4.0

    Practical software engineering in a terminal

    • DeepSeek R1 Qwen Abliterated: —
    • DeepSeek V4.1 Flash 1017 Thinking: 33.8%

  • SciCode

    Scientific computing in Python

    • DeepSeek R1 Qwen Abliterated: —
    • DeepSeek V4.1 Flash 1017 Thinking: 55.9%

Legacy evaluations

  • AA-LCR (unversioned)

    Long-context reasoning, earlier grading

    • DeepSeek R1 Qwen Abliterated: —
    • DeepSeek V4.1 Flash 1017 Thinking: 82.7%

Scores from Artificial Analysis, updated Oct 26, 2026. Artificial Analysis

Live routing data

Performance and reliability

Recent averages across requests routed for this model. Your own speed depends on prompt length, reasoning effort and load.

Throughput
15.6 tok/s
Faster than 7% of 258 models
Time to first token
12.4 s
Quicker than 5% of 258 models
Prompt cache hits
0.0%
Request success (24h)
58.0%

Cache hits are the share of input tokens recently served from a prompt cache across all requests. They are not a forecast of your own hit rate.

Developers

Use it through the API

Noumen’s API is OpenAI-compatible: any Chat Completions client works with a base URL and a key.

Read the quickstart
curl https://www.noumen.xyz/api/v1/chat/completions \
  -H "Authorization: Bearer $NOUMEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "huihui-ai/DeepSeek-R1-Distill-Qwen-32B-abliterated",
    "messages": [{ "role": "user", "content": "Explain this stack trace." }]
  }'
import os
from openai import OpenAI

client = OpenAI(base_url="https://www.noumen.xyz/api/v1", api_key=os.environ["NOUMEN_API_KEY"])
reply = client.chat.completions.create(
    model="huihui-ai/DeepSeek-R1-Distill-Qwen-32B-abliterated",
    messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)

DeepSeek

Other DeepSeek models, starting with the same family.