Dracarys Models by Abacus.AI

Abacus.AILlama 3 CommunityOpen weights
Abacus.AI releases · 3 in the last 12 monthsChangelog →
1 model2024Up to 128K ctx

Last refreshed 2026-05-19. Next refresh: weekly.

Details

ResearcherAbacus.AI
LicenseLlama 3 Community
Commercial useRestricted
Models1
Released2024
Max context128K

About

Dracarys is Abacus.AI's family of fine-tunes aimed at better coding performance, applied across a variety of base models and presented as the latest in its Smaug series. Dracarys-Llama-3.1-70B-Instruct is a fine-tune of Meta-Llama-3.1-70B-Instruct and keeps the same prompt format as Llama 3 70B Instruct. Abacus.AI's Hugging Face organization also lists Dracarys-72B-Instruct and Dracarys2 models. Weights are available on Hugging Face.

Decision facts

Best fit
coding
Capability starting point
Dracarys Llama 3.1 70B Instruct with 128K context
Lowest tracked input
Not tracked
Closest related family
Smaug 2

Current Variants

Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.

1 in view
Dracarys Llama 3.1 70B InstructCurrent

Use when the workload needs 128K context and 70B parameters.

2024-08128K context70B parameters

Release Timeline

1 release group
2024-08
1 current
Dracarys Llama 3.1 70B Instruct
128K context70B parameters
Current

Specifications(1 models)

Dracarys model specifications comparison
ModelReleasedContextParameters
Dracarys Llama 3.1 70B Instruct2024-08128K70B

Available From(1 provider)

NVIDIA NIM

Popular comparisons in this family