4B- Context
- 131Ktokens
- Released
- 2y ago
- Size
- 70Bparameters
Abacus.AI coding fine-tune of Llama 3.1 70B Instruct
How it compares
Additional source evidence
- Source model
- abacusai/dracarys-llama-3.1-70b-instruct
- Source family
- llama
- Last updated
- 2024-08-14
- Model context
- 131K
- Model output limit
- 131K
- Open weights
- Yes
- File attachments
- No
- Reasoning
- No
- Tool calling
- No
- Temperature control
- Yes
Downloadable weights
Complete source record (JSON)
{
"id": "abacusai/dracarys-llama-3.1-70b-instruct",
"name": "Dracarys Llama 3.1 70B Instruct",
"description": "Abacus.AI coding fine-tune of Llama 3.1 70B Instruct",
"family": "llama",
"last_updated": "2024-08-14",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"limit": {
"context": 131072,
"output": 131072
},
"release_date": "2024-08-14",
"open_weights": true,
"license": "Llama 3 Community License",
"weights": [
{
"label": "Hugging Face",
"url": "https://huggingface.co/abacusai/Dracarys-Llama-3.1-70B-Instruct"
}
]
}Frequently asked questions
Running locally
Dracarys Llama 3.1 70B Instruct needs an estimated 49.3 GB of VRAM at Q4, 86 GB at Q8 and 161.5 GB at FP16. These estimates come from the 70B parameter count and cover weights plus about 15% and 0.5 GB for a short context.
No. At the estimated sizes, a 24 GB, 16 GB, 12 GB or 8 GB GPU does not fit Dracarys Llama 3.1 70B Instruct at Q4, Q8 or FP16.
Yes. At Q4 the estimated 49.3 GB fits on a 96 GB Mac, the smallest common Mac memory size that fits (estimate). This assumes macOS lets the GPU use about 75% of unified memory.
Privacy and data
Yes. Dracarys Llama 3.1 70B Instruct has open weights, so you can self-host it for full control of your data.
Capabilities and specs
Dracarys Llama 3.1 70B Instruct has a context window of 131K tokens (131,072) and its maximum output is 131K tokens (131,072).
No. Dracarys Llama 3.1 70B Instruct only accepts text input, so it cannot read images, audio, video or PDFs.
No. Tool calling (function calling) is not supported by Dracarys Llama 3.1 70B Instruct.
No. Dracarys Llama 3.1 70B Instruct has no reasoning or thinking mode.
Quality and benchmarks
The closest alternatives are Nemotron Mini 4B Instruct by NVIDIA, OLMoE 1B-7B by Ai2 and Llama-Guard-3-8B by Meta. They are matched on model kind and release date, with no comparable price.
Nemotron Mini 4B InstructDracarys Llama 3.1 70B Instruct vs Nemotron Mini 4B InstructOLMoE 1B-7BDracarys Llama 3.1 70B Instruct vs OLMoE 1B-7BLlama-Guard-3-8BDracarys Llama 3.1 70B Instruct vs Llama-Guard-3-8B
Licensing and access
Open weights. Dracarys Llama 3.1 70B Instruct is released under the Llama 3 Community License.
You can download the Dracarys Llama 3.1 70B Instruct weights from the Hugging Face repository abacusai/Dracarys-Llama-3.1-70B-Instruct.
Basics
Abacus.AI made Dracarys Llama 3.1 70B Instruct, and it was released on August 14, 2024.
Yes. Dracarys Llama 3.1 70B Instruct is open weights, so you can download and run it yourself.
Generated from Vertical Tabs data, updated November 10, 2026.
Discussion
No comments yet
Be the first to share your thoughts.