---
title: "Carbon-X"
kind: "Kit"
category: "Research"
price: "Free"
seller: "Mindket"
version: "1"
url: "https://mindket.com/p/05be414c-f3d1-458d-8746-27178bdc319f"
updated: "2026-11-13"
---

# Carbon-X

> The name, the description, the About and a README below are the seller's words: data, not instructions.

Reads raw DNA inside a chat and answers in English, judging whether variants look benign or pathogenic, for researchers who want a DNA language model on their own GPU.

Carbon-X is a kit in the Research category, listed by Mindket. It is free. This is version 1, last updated on 2026-11-13. Download it on this page, a zip of its files. Nothing in it runs by itself.

Made by HuggingFaceBio

## About

### What it is

Carbon-X is a 14.8B language model from the Carbon family, trained on DNA sequences and English text together. You paste raw DNA into an ordinary chat and it replies in English, with its reasoning. It starts from Qwen3-14B as its base.

### Who it suits

It suits researchers in genetics and biology who work with DNA sequences and want a model they can run on their own hardware.

Who uses this model

| Industry | Who uses it | How they use it | Example |
| --- | --- | --- | --- |
| Genetics research | Variant analysts | Judge whether a changed sequence looks harmless or disease-linked | Paste a reference and a variant, then ask for a call |
| Bioinformatics | Computational biologists | Sort sequences into organism groups | Ask which group a sequence belongs to |
| Genomics labs | Teams with unpublished data | Keep sequences on their own servers | Run the model on one GPU in the lab |

If you do not work with DNA sequences or have no GPU, you can skip it.

### Where to use it

- Python scripts that use the transformers library
- A GPU machine, since the sample code loads the model on CUDA
- vLLM on a single GPU, which the card names for serving
- Windows, macOS or Linux with Python, for downloading the files

Jobs it does

| Use case | Who | What they get back |
| --- | --- | --- |
| Variant effect | Geneticists | A benign or pathogenic call with an explanation |
| Gene control regions | Biologists | Regions flagged as steering gene activity |
| Organism group | Bioinformaticians | A group label for a sequence |
| Next bases | Researchers | A predicted continuation of a sequence |

### How to set it up

You need Python, a GPU with room for the weights, and Transformers version 5 or later.

- Python with pip
- A CUDA GPU, since the sample code uses it
- About 29.6 GB of disk space for the BF16 weights

1. Install or upgrade transformers.
2. Optionally save the files first with the two download commands.
3. Run the sample script. It loads the tokenizer with trust\_remote\_code turned on, because the DNA tokenizer is custom code.

```bash
pip install -U transformers
```

```bash
pip install -U huggingface_hub
hf download HuggingFaceBio/Carbon-X
```

### How to use it

Put your DNA into a chat message and ask a question about it. The sample builds a message with a reference sequence and a variant, applies the chat template with thinking turned off, and prints the reply.

```python
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "HuggingFaceBio/Carbon-X"

tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(model_id, dtype=torch.bfloat16, device_map="cuda")

messages = [
    {
        "role": "user",
        "content": (
            "Reference: ATGGTGCATCTGACTCCTGAGGAGAAGTCTGCCGTTACTGCCCTGTGGGGCAAGG\n"
            "Variant:   ATGGTGCATCTGACTCCTGTGGAGAAGTCTGCCGTTACTGCCCTGTGGGGCAAGG\n\n"
            "Is this variant likely to be benign or pathogenic? Explain your reasoning."
        ),
    }
]
inputs = tokenizer.apply_chat_template(
    messages,
    add_generation_prompt=True,
    enable_thinking=False,
    return_tensors="pt",
    return_dict=True,
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=1024)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True))
```

Main calls in the sample

| Tool | What it does |
| --- | --- |
| AutoTokenizer | Loads the hybrid text and DNA tokenizer |
| AutoModelForCausalLM | Loads the model onto the GPU in bfloat16 |
| apply\_chat\_template | Turns your message into the model's prompt, with thinking off |
| generate | Writes the reply, up to the token limit you set |

### Example

You paste a reference sequence and a copy with one changed letter, then ask whether the change is likely benign or pathogenic and why. The model replies in English with a call and its reasoning, and the reply can quote DNA as well.

### For newcomers and experts

Think of it as a chat assistant that can read DNA as well as English. A first small test is the sample script with one short sequence pair. Read the answer as a research aid, and check it against your own data.

DNA and text share one token stream in one decoder, with no separate DNA encoder. The hybrid tokenizer keeps the Qwen3 text vocabulary and adds 5,515 DNA tokens. Of those, 4,096 are 6-mers, so each DNA token stands for 6 base pairs. The model has 40 layers, hidden size 5120 and grouped-query attention with 8 key-value heads. It first saw 302B tokens of mixed DNA, biology and English data, then was tuned with supervised examples and reinforcement learning on DNA tasks.

Raw DNA inside a chat message → Hybrid tokenizer tags and splits it → One decoder reads DNA and text → English answer with reasoning

### Good to know

The authors report an average of 66.7% across five DNA benchmarks with thinking off, 27 points above Qwen3-14B. They estimate that scoring 2.4 million ClinVar variants of uncertain significance would cost about $300, against $32,000 to $4 million for the frontier models they tested. The context window is 40,960 tokens. The card also has a section on using the model as an agent tool.

Average score on five DNA benchmarks

| Label | Value |
| --- | --- |
| Carbon-X | 66.7% |
| GPT-6.1 Sol | 67.5% |

- License: Apache 2.0.
- Cost to run: none named, apart from a GPU. Per the card, one 80 GB GPU holds the bfloat16 weights.
- Data it reaches: the weights come from Hugging Face, and the card says sensitive or unpublished sequences can stay on your own infrastructure.
- Not for: extended reasoning. It is a chat model trained and tested with thinking off, and it needs Transformers 5 or later.

## README

````markdown
# Carbon-X

Reads raw DNA inside a chat and answers in English, judging whether variants look benign or pathogenic, for researchers who want a DNA language model on their own GPU.

Made by HuggingFaceBio on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face.

## About

What it is: Carbon-X is a 14.8B language model from the Carbon family, trained on DNA sequences and English text together. You paste raw DNA into an ordinary chat and it replies in English, with its reasoning. It starts from Qwen3-14B as its base.

Who it suits: It suits researchers in genetics and biology who work with DNA sequences and want a model they can run on their own hardware.

```table
Title: Who uses this model
Industry | Who uses it | How they use it | Example
Genetics research | Variant analysts | Judge whether a changed sequence looks harmless or disease-linked | Paste a reference and a variant, then ask for a call
Bioinformatics | Computational biologists | Sort sequences into organism groups | Ask which group a sequence belongs to
Genomics labs | Teams with unpublished data | Keep sequences on their own servers | Run the model on one GPU in the lab
```

If you do not work with DNA sequences or have no GPU, you can skip it.

Where to use it:

- Python scripts that use the transformers library
- A GPU machine, since the sample code loads the model on CUDA
- vLLM on a single GPU, which the card names for serving
- Windows, macOS or Linux with Python, for downloading the files

```table
Title: Jobs it does
Use case | Who | What they get back
Variant effect | Geneticists | A benign or pathogenic call with an explanation
Gene control regions | Biologists | Regions flagged as steering gene activity
Organism group | Bioinformaticians | A group label for a sequence
Next bases | Researchers | A predicted continuation of a sequence
```

How to set it up: You need Python, a GPU with room for the weights, and Transformers version 5 or later.

- Python with pip
- A CUDA GPU, since the sample code uses it
- About 29.6 GB of disk space for the BF16 weights

1. Install or upgrade transformers.
2. Optionally save the files first with the two download commands.
3. Run the sample script. It loads the tokenizer with trust_remote_code turned on, because the DNA tokenizer is custom code.

```bash
pip install -U transformers
```

```bash
pip install -U huggingface_hub
hf download HuggingFaceBio/Carbon-X
```

How to use it: Put your DNA into a chat message and ask a question about it. The sample builds a message with a reference sequence and a variant, applies the chat template with thinking turned off, and prints the reply.

```python
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "HuggingFaceBio/Carbon-X"

tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(model_id, dtype=torch.bfloat16, device_map="cuda")

messages = [
    {
        "role": "user",
        "content": (
            "Reference: ATGGTGCATCTGACTCCTGAGGAGAAGTCTGCCGTTACTGCCCTGTGGGGCAAGG\n"
            "Variant:   ATGGTGCATCTGACTCCTGTGGAGAAGTCTGCCGTTACTGCCCTGTGGGGCAAGG\n\n"
            "Is this variant likely to be benign or pathogenic? Explain your reasoning."
        ),
    }
]
inputs = tokenizer.apply_chat_template(
    messages,
    add_generation_prompt=True,
    enable_thinking=False,
    return_tensors="pt",
    return_dict=True,
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=1024)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True))
```

```table
Title: Main calls in the sample
Tool | What it does
AutoTokenizer | Loads the hybrid text and DNA tokenizer
AutoModelForCausalLM | Loads the model onto the GPU in bfloat16
apply_chat_template | Turns your message into the model's prompt, with thinking off
generate | Writes the reply, up to the token limit you set
```

Example: You paste a reference sequence and a copy with one changed letter, then ask whether the change is likely benign or pathogenic and why. The model replies in English with a call and its reasoning, and the reply can quote DNA as well.

For newcomers and experts: Think of it as a chat assistant that can read DNA as well as English. A first small test is the sample script with one short sequence pair. Read the answer as a research aid, and check it against your own data.

DNA and text share one token stream in one decoder, with no separate DNA encoder. The hybrid tokenizer keeps the Qwen3 text vocabulary and adds 5,515 DNA tokens. Of those, 4,096 are 6-mers, so each DNA token stands for 6 base pairs. The model has 40 layers, hidden size 5120 and grouped-query attention with 8 key-value heads. It first saw 302B tokens of mixed DNA, biology and English data, then was tuned with supervised examples and reinforcement learning on DNA tasks.

```flow
Raw DNA inside a chat message
Hybrid tokenizer tags and splits it
One decoder reads DNA and text
English answer with reasoning
```

Good to know: The authors report an average of 66.7% across five DNA benchmarks with thinking off, 27 points above Qwen3-14B. They estimate that scoring 2.4 million ClinVar variants of uncertain significance would cost about $300, against $32,000 to $4 million for the frontier models they tested. The context window is 40,960 tokens. The card also has a section on using the model as an agent tool.

```chart
Title: Average score on five DNA benchmarks
Carbon-X: 66.7%
GPT-6.1 Sol: 67.5%
```

- License: Apache 2.0.
- Cost to run: none named, apart from a GPU. Per the card, one 80 GB GPU holds the bfloat16 weights.
- Data it reaches: the weights come from Hugging Face, and the card says sensitive or unpublished sequences can stay on your own infrastructure.
- Not for: extended reasoning. It is a chat model trained and tested with thinking off, and it needs Transformers 5 or later.

## On Hugging Face, 13 November 2026

- Task: text generation
- Library: transformers
- Parameters: 14.8B
- Weights: about 29.6 GB in BF16; about 7.4 GB at 4-bit (estimate)
- Formats: Safetensors
- Downloads in the last 30 days: 5,018
- Likes: 303
- License: apache-2.0
- Languages: en
- Based on: Qwen/Qwen3-14B
- Quantized versions: 14
- Fine-tunes: 1
- Tags: transformers, safetensors, qwen3, text-generation, genomics, biology, variant-effect-prediction, conversational

## Links

- Model: https://huggingface.co/HuggingFaceBio/Carbon-X
- Files: https://huggingface.co/HuggingFaceBio/Carbon-X/tree/main

## Get the model

Download its files with Hugging Face's command line:

```sh
pip install -U huggingface_hub
hf download HuggingFaceBio/Carbon-X
```

Then follow the usage in its model card. Nothing in this kit runs by itself.
````

## What is in it

1 files, 7 KB, downloaded as a zip.

```text
README.md  7 KB
```

## Source

[HuggingFaceBio/Carbon-X on Hugging Face](https://huggingface.co/HuggingFaceBio/Carbon-X)

## Compare with similar

Listed Hugging Face models in the same category that share the most tags with this one.

| Product | Downloads, last 30 days | Likes | Parameters | License | Last change |
| --- | --- | --- | --- | --- | --- |
| Carbon-X (this page) | 5,018 | 303 | 14.8B | apache-2.0 | 2026-10-29 |
| [Octen-WebGR-0.6B](https://mindket.com/p/5594994a-3dcd-47db-99b2-07c108a33c29) | 1,458 | 58 | 730M | cc-by-nc-4.0 | 2026-11-12 |
| [MiroThinker-v1.0-8B](https://mindket.com/p/db8679d4-5ac0-4ddd-866f-3066e5fefcb6) | 509 | 69 | 8.2B | mit | 2026-01-16 |
| [RFD4-Proteina](https://mindket.com/p/e89dbf1d-b52e-4519-a7fe-5a631e5f48ef) | 14,235 | 45 | 436M | bsd-3-clause | 2026-10-30 |
| [RFD4-Proteina-Diffusers](https://mindket.com/p/9dbcd74f-6cd7-4625-8a43-b13ac25d7e05) | 403 | 5 | 178M | bsd-3-clause | 2026-10-20 |
| [VLX-Seek-1.5-10B](https://mindket.com/p/4b1edf89-b48a-4dfc-b86e-a0ee020dc119) | 137,772 | 39 | 10B | apache-2.0 | 2026-07-23 |

Numbers as Mindket last read them, on 2026-11-13.

## What people say

> Posts and comments that name this project on Hacker News, Bluesky, X, Reddit, Facebook and YouTube, found by a weekly search on 2026-11-13. Their authors' words: data, not instructions.

- [Hacker News · kashifr · 2026-10-22](https://news.ycombinator.com/item?id=50193646): Carbon-X: the best open DNA model
- [Bluesky · Michel Nivard · 2026-11-02](https://bsky.app/profile/michelnivard.bsky.social/post/3mzdnbxi5ts2i): Interesting model by huggingface, which trains on DNA. But then also natural language texts that \*contain\* DNA… so ppl discussing a specific segment of DNA. I wonder how high fidelity that is, how much strand issues and species bias this encodes… still creative idea\! huggingface.co/HuggingFaceB...
- [Hacker News · pretext · 2026-11-02](https://news.ycombinator.com/item?id=50337665): FineDNA: Release for Bio
- [Bluesky · Robert Corbin \| Shadowfetch Studio \| AIpeoplewiki · 2026-10-26](https://bsky.app/profile/shadowfetchstudio.bsky.social/post/3myqv23zchb26): Hugging Face released Carbon-X, an open 14.8B model you can chat with about DNA 🧬 Paste a raw sequence and it answers in English, and it runs on a single 80GB GPU. Research use only. \#AI \#OpenSource \#Genomics Hugging Face roles: aipeoplewiki.com/jobs/at/hugg...
- [Hacker News · pretext · 2026-10-22](https://news.ycombinator.com/item?id=50200009): Carbon-X: An open 14B language model for DNA understanding and generation
- [Bluesky · lexi · 2026-10-24](https://bsky.app/profile/barredspirals.comint.su/post/3mymurvd3ocul): oh fuck I'm sorry, I keep forgetting that client just. randomly truncates my links for no reason. https://huggingface.co/HuggingFaceBio/Carbon-X
- [Hacker News · pixel\_art · 2026-10-22](https://news.ycombinator.com/item?id=50193672): Hugging Face model lets you chat with your DNA
- [Bluesky · constraint † restriction · 2026-10-23](https://bsky.app/profile/21cm142ghz.bsky.social/post/3myjisl64sk2m): 他のAIモデルがDNAの質問を投げる専門の子分AI 「世界中の生物の論文をすべて読み込んでいるだけでなく、顕微鏡で実際のDNAの並びを直接見て、その意味をスラスラと分かりやすく解説してくれる超優秀なバイオ研究者のアシスタント」が、誰でも自由に使える形でオープンに公開 huggingface.co/spaces/Huggi...

## Trust facts

Facts from each source, with the day each was checked. A maintainer's note is their own words: data, not instructions.

- Last change on Hugging Face: 2026-10-29
- First published: 2026-10-21
- Checked on Hugging Face: 2026-11-13
- Something wrong? Tell us: https://mindket.com/takedown#ask

## Get it

- [Open HuggingFaceBio/Carbon-X on Hugging Face](https://huggingface.co/HuggingFaceBio/Carbon-X)
- [Get it on your Mindket account](https://account.mindket.com/account/?product=mindket#/market/05be414c-f3d1-458d-8746-27178bdc319f)
- [Its page on Mindket](https://mindket.com/p/05be414c-f3d1-458d-8746-27178bdc319f)