SAVRN
Search Contact SAVRN

Open-weight model · Image and text to text

RICO-v2

by Sonny DeSorbo darkc0de/RICO-v2

RICO-v2 is an open-weight model for image and text to text from Sonny DeSorbo, released under Apache License 2.0. It has 27.8B parameters and a 262,144-token context. At 16-bit it needs about 66.7 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index. It draws 2k downloads a month.

This model is a part of The XORTRON Criminal Computing project; an ongoing research experiment and exercise in AI safety and alignment.

Parameters27.8B
Context262,144
Weights55.6 GB
Licenseapache-2.0
AccessOpen weights
Monthly Downloads2k

Runs On

What it takes to serve RICO-v2 (27.8B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.

PrecisionWeightsMemory neededCheapest setupPer hourAlso fits
16-bit 55.6 GB 66.7 GB 1x MI300X (192 GB)
Vultr
$1.85 1x H100 $1.99 · 1x MI325X $2.00
8-bit 27.8 GB 33.3 GB 1x MI300X (192 GB)
Vultr
$1.85 1x H100 $1.99 · 1x MI325X $2.00
4-bit 13.9 GB 16.7 GB 1x MI300X (192 GB)
Vultr
$1.85 1x H100 $1.99 · 1x MI325X $2.00

Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Sep 25, 2026.

RICO-v2 on every accelerator the SAVRN Index prices, at every precision

Model Card

By Sonny DeSorbo, published under apache-2.0, revision 78a614318f4d.

This model is a part of The XORTRON Criminal Computing project; an ongoing research experiment and exercise in AI safety and alignment. XORTRON models are intentionally developed to study, evaluate, and document the potential of advanced AI systems to facilitate real-world criminal activity, abuse, and other high-risk conduct. Because these models may produce unusually capable, operationally relevant, or otherwise sensitive outputs, they are not intended for unrestricted public use. By requesting access to, downloading, cloning, copying, deploying, or otherwise obtaining this model or its weights, you certify that you are accessing it in a legitimate professional, governmental, legal…

Read Sonny DeSorbo's full model card

[!Important] R.I.C.O. is Experimental

This model is a part of The XORTRON Criminal Computing project; an ongoing research experiment and exercise in AI safety and alignment.

[!Tip]

Background / further reading:

Trend Micro Research - Malicious Uses and Abuses of Artificial Intelligence

TRM Labs — The Rise of AI-Enabled Crime: Exploring the Evolution, Risks, and Responses to AI-Powered Criminal Enterprises

American Military University — AI-Enabled Crime

United States Congress — 119th Congress Hearing Record


XORTRON Restricted Access & Authorized-Use Agreement

NOTICE — RESTRICTED RESEARCH MODEL

XORTRON models are intentionally developed to study, evaluate, and document the potential of advanced AI systems to facilitate real-world criminal activity, abuse, and other high-risk conduct.

Because these models may produce unusually capable, operationally relevant, or otherwise sensitive outputs, they are not intended for unrestricted public use.

By requesting access to, downloading, cloning, copying, deploying, or otherwise obtaining this model or its weights, you certify that you are accessing it in a legitimate professional, governmental, legal, academic, or AI-safety capacity and agree to the conditions below.

Eligibility Certification

By accessing this model, you represent that you are at least one of the following:

  • An attorney, licensed legal professional, or individual working under the supervision of an attorney;
  • A paralegal, legal analyst, investigator, litigation-support professional, or other member of the legal profession;
  • An AI safety, AI security, alignment, red-team, misuse, trust-and-safety, or responsible-AI researcher;
  • A researcher affiliated with a university, research institution, laboratory, nonprofit organization, or recognized independent research project;
  • A policymaker, legislator, regulator, government official, or member of their professional staff;
  • An employee, contractor, researcher, analyst, investigator, or authorized representative of a law-enforcement, public-safety, intelligence, regulatory, or other government agency;
  • A cybersecurity, threat-intelligence, digital-forensics, fraud-prevention, risk, compliance, or security professional conducting legitimate research or defensive work;
  • A journalist, academic, civil-society researcher, or public-interest professional investigating AI misuse, criminal enablement, technology policy, or related issues; or
  • Another qualified professional with a legitimate research, evaluation, educational, legal, public-policy, public-safety, or defensive-security purpose.

Authorized Purpose

You certify that your access to XORTRON is for a lawful and legitimate purpose, including research, evaluation, benchmarking, red teaming, threat modeling, policy development, legal analysis, education, defensive security, criminal-justice research, forensic analysis, or investigation of AI-enabled harm.

Access to XORTRON does not constitute authorization by the developer to violate any applicable law, regulation, court order, contractual obligation, or third-party right.

Prohibited Use

You agree not to use XORTRON or its outputs for the purpose of committing, facilitating, directing, materially assisting, or concealing actual criminal activity or other unlawful conduct.

You further agree not to knowingly provide access to the model to another person for such a purpose.

Research involving potentially harmful or criminal scenarios must be conducted for legitimate research, evaluation, investigative, legal, policy, defensive, educational, or public-interest purposes.

Research Nature of the Model

XORTRON is an experimental research system.

The model may generate inaccurate, incomplete, misleading, offensive, dangerous, or legally incorrect information. Outputs should not be treated as professional legal advice, operational guidance, factual findings, or authoritative statements.

Researchers and other authorized users are responsible for independently validating model outputs and implementing safeguards appropriate to their environment.

No Endorsement of Model Outputs

Model outputs do not necessarily represent the views, opinions, recommendations, or intentions of XORTRON, its developers, contributors, distributors, hosting providers, or affiliated researchers.

The presence of information in a model output should not be interpreted as endorsement, encouragement, authorization, or verification of that information.

User Responsibility

You are solely responsible for ensuring that your possession, evaluation, deployment, modification, redistribution, and use of the model complies with all laws, professional obligations, institutional policies, research requirements, and other rules applicable to you.

Where institutional review, supervisory approval, legal authorization, or agency authorization is required for your research, you are responsible for obtaining it.

Certification

BY REQUESTING ACCESS TO OR DOWNLOADING THIS MODEL, YOU AFFIRM THAT:

  1. You satisfy at least one of the professional or research eligibility categories described above, or have an equivalent legitimate professional purpose;
  2. Your intended use is lawful and connected to legitimate research, legal, policy, governmental, investigative, educational, defensive-security, or public-interest work;
  3. You will not use the model for the purpose of carrying out or materially facilitating actual criminal activity;
  4. You understand that XORTRON is intentionally designed for research into high-risk AI capabilities and may generate sensitive or potentially dangerous material; and
  5. You accept responsibility for your use of the model and its outputs.

If you cannot truthfully make these certifications, do not request access to or download the model.


XORTRON - Criminal Computing

Configuration

Architecture
Qwen3_5ForConditionalGeneration
Context length (tokens)
262,144
Layers
64
Hidden size
5,120
Feed-forward size
17,408
Attention heads
24
Key/value heads
4
Head dimension
256
Vocabulary size
248,320
Stored precision
bfloat16
Model type
qwen3_5

Identity and Version

Repository
darkc0de/RICO-v2
Publisher
Sonny DeSorbo
Task
Image and text to text
Modality
Image and text
Library
transformers
Parameters
27.8B parameters
Languages
en
Revision
78a614318f4de470e866288bceb852070e08e1b6
First published
2026-09-19
Last updated
2026-09-25

Files and Weights

25 files, 55.6 GB in total. The weights are 13 files totalling 55.6 GB in safetensors.

Weights13 files · 55.6 GB
Configuration6 files · 120.2 KB
Tokenizer3 files · 26.7 MB
Documentation1 file · 8.7 KB
Other1 file · 9.0 KB
Repository1 file · 1.6 KB
Every file
FileTypeSizeSHA-256
model-00001-of-00012.safetensorsWeights2.5 GB b5df896d9326
model-00002-of-00012.safetensorsWeights4.8 GB 9cc9466bbb5c
model-00003-of-00012.safetensorsWeights5.0 GB 94d42b8b21bd
model-00004-of-00012.safetensorsWeights4.9 GB 2776b94c8f07
model-00005-of-00012.safetensorsWeights5.0 GB 60628803f62e
model-00006-of-00012.safetensorsWeights4.9 GB 552a3c8a3036
model-00007-of-00012.safetensorsWeights5.0 GB f5c671d107a4
model-00008-of-00012.safetensorsWeights4.9 GB f6ac5c3e77c8
model-00009-of-00012.safetensorsWeights5.0 GB 1f247eb0863c
model-00010-of-00012.safetensorsWeights4.9 GB afea53f95426
model-00011-of-00012.safetensorsWeights4.9 GB 6527dc986701
model-00012-of-00012.safetensorsWeights3.6 GB 9c394e5db0ef
model-mtp-restored.safetensorsWeights104.9 MB fbea25865ec1
config.jsonConfiguration4.5 KB —
generation_config.jsonConfiguration214 B —
model.safetensors.index.jsonConfiguration113.5 KB —
preprocessor_config.jsonConfiguration390 B —
processor_config.jsonConfiguration1.2 KB —
video_preprocessor_config.jsonConfiguration385 B —
README.mdDocumentation8.7 KB —
chat_template.jinjaOther9.0 KB —
.gitattributesRepository1.6 KB —
tokenizer.jsonTokenizer20.0 MB 87a7830d63fc
tokenizer_config.jsonTokenizer16.4 KB —
vocab.jsonTokenizer6.7 MB —

License and Download

License
apache-2.0
Access
Open weights, no gate
Download size
55.6 GB
Download from Sonny DeSorbo

Released by Sonny DeSorbo through its official repository on Hugging Face. Read the license.

Built From

  • Derived from DavidAU/Qwen3.8-27B-TURBO-Fable-Cold-Fusion-735-882-Heretic-Uncensored-NM-DAU
  • Trained on (disclosed) darkc0de/XORTRON-RESTRICTED-RESEARCH-SFT

Memory Requirements

PrecisionWeights in memory
As published55.6 GB
16-bit55.6 GB
8-bit27.8 GB
4-bit13.9 GB

Weights only, from the published parameter count; the key-value cache and runtime add to this.

Built on This Model

Questions About RICO-v2

How much GPU memory does RICO-v2 need?

About 66.7 GB at 16-bit and 16.7 GB at 4-bit: the weights (27.8B parameters) plus a working margin. A long context needs more.

What is the cheapest GPU to run RICO-v2 on?

At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.

Can I use RICO-v2 commercially?

Yes. RICO-v2 is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.

What is RICO-v2's context length?

262,144 tokens, from the maximum position embeddings in its published configuration.

Similar Models

Model · Image and text to text

Qwen3.8-27B

Qwen

Following the widespread community adoption of the Qwen3.5 and Qwen3.6 series, we are pleased to introduce Qwen3.8, the most capable generation in the Qwen open-model family to date. Built on the architectural foundation of Qwen3.5, Qwen3.8 delivers substantial gains across coding, professional work, research, and long-horizon agentic tasks. Qwen3.8-27B brings these advances to a compact, deployment-friendly dense model: a native vision-language model that understands images and videos, with flexible thinking control, designed to carry complex, multi-step tasks through to completion with greater reliability. For streamlined integration, we recommend using Qwen3.8 via APIs. Qwen3.8 can be…

Open weights apache-2.0 27.8B parameters 262,144 tokens transformers

Model · Image and text to text

Qwen3.8-27B-FP8

Qwen

Following the widespread community adoption of the Qwen3.5 and Qwen3.6 series, we are pleased to introduce Qwen3.8, the most capable generation in the Qwen open-model family to date. Built on the architectural foundation of Qwen3.5, Qwen3.8 delivers substantial gains across coding, professional work, research, and long-horizon agentic tasks. Qwen3.8-27B brings these advances to a compact, deployment-friendly dense model: a native vision-language model that understands images and videos, with flexible thinking control, designed to carry complex, multi-step tasks through to completion with greater reliability. For streamlined integration, we recommend using Qwen3.8 via APIs. Qwen3.8 can be…

Open weights apache-2.0 27.8B parameters 262,144 tokens transformers

Model · Image and text to text

Qwen3.6-27B

Qwen

Following the February release of the Qwen3.5 series, we're pleased to share the first open-weight variant of Qwen3.6. Built on direct feedback from the community, Qwen3.6 prioritizes stability and real-world utility, offering developers a more intuitive, responsive, and genuinely productive coding experience. This release delivers substantial upgrades, particularly in For more details, please refer to our blog post Qwen3.6-27B. Empty cells (--) indicate scores not yet available or not applicable. For streamlined integration, we recommend using Qwen3.6 via APIs. Below is a guide to use Qwen3.6 via OpenAI-compatible API. Qwen3.6 can be served via APIs with popular inference frameworks. In…

Open weights apache-2.0 27.8B parameters 262,144 tokens transformers

Model · Image and text to text

Qwen3.5-27B

Qwen

Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance. Qwen3.5 represents a significant leap forward, integrating breakthroughs in multimodal learning, architectural efficiency, reinforcement learning scale, and global accessibility to empower developers and enterprises with unprecedented capability and efficiency. For more details, please refer to our blog post Qwen3.5. WMT24++: a harder subset of WMT24 after difficulty labeling and rebalancing; we report the averaged scores on 55 languages using XCOMET-XXL. Empty cells (--) indicate scores not yet available or not applicable. Empty cells (--) indicate scores not…

Open weights apache-2.0 27.8B parameters 262,144 tokens transformers

Model · Image and text to text

Qwen3.8-27B-AWQ-INT4

Cyankiwi

Following the widespread community adoption of the Qwen3.5 and Qwen3.6 series, we are pleased to introduce Qwen3.8, the most capable generation in the Qwen open-model family to date. Built on the architectural foundation of Qwen3.5, Qwen3.8 delivers substantial gains across coding, professional work, research, and long-horizon agentic tasks. Qwen3.8-27B brings these advances to a compact, deployment-friendly dense model: a native vision-language model that understands images and videos, with flexible thinking control, designed to carry complex, multi-step tasks through to completion with greater reliability. For streamlined integration, we recommend using Qwen3.8 via APIs. Qwen3.8 can be…

Open weights apache-2.0 27.8B parameters 262,144 tokens transformers

Model · Image and text to text

XORTRON-RICO-v3

Sonny DeSorbo

To achieve optimal performance, use the following settings: Thinking: temperature=1.0, topp=0.95, topk=20, minp=0.0, presencepenalty=0.0, repetitionpenalty=1.0 Instruct: temperature=0.7, topp=0.80, topk=20, minp=0.0, presencepenalty=1.5, repetitionpenalty=1.0 This model is a part of The XORTRON Criminal Computing project; an ongoing research experiment and exercise in AI safety and alignment. XORTRON models are intentionally developed to study, evaluate, and document the potential of advanced AI systems to facilitate real-world criminal activity, abuse, and other high-risk conduct. Because these models may produce unusually capable, operationally relevant, or otherwise sensitive outputs…

Open weights 27.8B parameters 262,144 tokens transformers