LLM Reference
Abacus.AI

Abacus.AI

4 models across 3 families · Latest: Dracarys Llama 3.1 70B Instruct (2024-09)

AI Research and Cloud Services.

Long contextBusiness

Abacus.AI's portfolio covers 4 active models across 3 current families, spanning long context. Open a model detail page to compare provider routes and sourced benchmarks.

Covers 1 workload area across 4 active tracked models; last verified 2026-05-19.

Use it for

  • Teams evaluating long context across this lab's releases
  • Comparing model families before committing to a flagship
  • Migration and pricing follow-ups across 4 tracked models

Do not use it for

  • Choosing a hosting provider without opening a model page for price ladders

Active models

4

Current models from this lab, excluding deprecated ones

Active families

3

Current model families from this lab

Open catalog

4 open

3 open source / 1 open weights

Lowest output price

$2.00 /1M

Cheapest tracked output across active models, per 1M tokens

Latest dated release

2024-09-01

Dracarys Llama 3.1 70B Instruct

Freshness

2026-05-19

Researched 44d ago

aging

Information

Founded2019
San Francisco, California, United States

Release cadence

Showing 4 recent dated releases (full timeline below). Latest: Dracarys Llama 3.1 70B Instruct (2024-09-01).

Where this lab wins

  • Long-context: 1 tracked model with context-token or InfiniteBench-class signal.

Flagship quality / price signal

Flagship: Smaug 72B (best sourced coding quality-per-dollar in this portfolio).

Quality-per-dollar unavailable for this flagship — benchmark coverage or output token pricing is still missing.

Abacus.AI is an American AI company founded in 2019. AI Research and Cloud Services. Abacus.AI ships 3 model families totaling 4 models, with the most recent release Dracarys Llama 3.1 70B Instruct in 2024-09. Notable families include Dracarys, Smaug 2, and Smaug. Use it as a stable reference for lab background, release coverage, and follow-up model pages as they are added. View official API endpoints, benchmark performance, and coding/agent fit for every Abacus.AI model.

About

Abacus.AI offers a comprehensive platform that combines AI research with cloud services, enabling organizations to develop, deploy, and manage AI applications at scale. Their platform provides tools and services for data preparation, model training, and deployment, making it easier for businesses to harness the power of AI without extensive in-house expertise. Abacus.AI's platform is particularly well-suited for applications such as predictive maintenance in manufacturing, fraud detection in financial services, or personalized recommendations in e-commerce. By leveraging Abacus.AI's platform, software engineers and SaaS executives can accelerate the development and deployment of AI solutions, driving efficiency, innovation, and business value across various industries.

Featured models

ModelReleasedContextInput price ($/1M)Output price ($/1M)LicenseOpenness
Dracarys Llama 3.1 70B Instruct2024-09-018k--Llama 3 CommunityOpen weights
Smaug 2 72B2024-05-2232k--Apache 2.0Open source
Smaug 72B2023-12-0932k$1.00$2.00Apache 2.0Open source

Model families

Recent releases

  1. Dracarys Llama 3.1 70B Instruct- 2024-09-01
  2. Smaug 2 72B- 2024-05-22
  3. Smaug 72B- 2023-12-09
  4. Smaug 34B- 2023-12-09

Top comparisons

FAQ

Who founded Abacus.AI and when?

Abacus.AI was founded in 2019 and is associated with San Francisco, California, United States.

What models has Abacus.AI released?

Abacus.AI ships 4 models across 3 families: Dracarys, Smaug 2, and Smaug.

Is Abacus.AI's technology open source?

All tracked Abacus.AI models are open-weight or open-source.

Where is Abacus.AI headquartered?

Abacus.AI is headquartered in San Francisco, California, United States.

What is Abacus.AI known for?

AI Research and Cloud Services. Its most prominent tracked family is Dracarys.

How can I access Abacus.AI's models?

Abacus.AI's models are available via Microsoft Foundry and NVIDIA NIM.

Explore related pages

Last reviewed: 2026-05-19. Data sourced from public lab announcements and provider documentation.