open-source

TII (Falcon)

Abu Dhabi research lab that pioneered permissively-licensed frontier open-source models.

4
Models tracked
2020
Founded
Falcon 180B
Flagship model
Abu Dhabi
HQ

About

About TII (Falcon)

The Technology Innovation Institute (TII) in Abu Dhabi made history by releasing Falcon 40B under a fully permissive Apache 2.0 licence — the first frontier-scale model with true commercial freedom. The Falcon series continued with Falcon 180B (the largest openly available dense model at its release), Falcon 2 with native vision support, and Falcon-Mamba 7B which introduced Mamba state-space architecture to mainstream open-source AI.

Key highlights

  • First lab to release a frontier-scale model (40B) under Apache 2.0 licence

  • Falcon 180B was the largest publicly available dense model at its release

  • Falcon-Mamba 7B introduces Mamba SSM with linear inference scaling to 128K

  • Falcon 2 adds vision-language capability with full commercial use rights

  • All models freely available on Hugging Face with no usage restrictions

Models

4 models tracked

Context windows, pricing, capabilities, and API identifiers — everything you need to choose the right model for your workload.

FlagshipOpen Source

Falcon 180B

2023

2Kcontext
TextCode

The largest dense model ever released under a fully permissive open-source licence at time of release. 180B parameters trained on 3.5 trillion tokens with full commercial rights.

  • Largest permissive open model at launch
  • Apache 2.0 licence
  • High quality ceiling
  • Full commercial freedom
tiiuae/falcon-180B-chat
Open Source

Falcon 2 11B

2024

8Kcontext
TextVisionCode

The second generation Falcon model with native vision-language capability. Handles image understanding and VQA tasks with full commercial-use permissions.

  • Vision-language capability
  • 8K context
  • Apache 2.0
  • Efficient 11B size
tiiuae/falcon-11B
Open Source

Falcon-Mamba 7B

2024

128Kcontext
Text

A pure Mamba state-space model that processes 128K tokens with linear (not quadratic) inference cost. Ideal for very long sequences where attention overhead becomes prohibitive.

  • Pure Mamba SSM architecture
  • Linear inference cost
  • 128K context
  • No attention overhead
tiiuae/falcon-mamba-7b-instruct
Open Source

Falcon 40B

2023

2Kcontext
Text

The original model that proved permissive frontier models were possible. Falcon 40B under Apache 2.0 was a watershed moment for the open-source AI ecosystem.

  • Apache 2.0 original pioneer
  • Proven quality
  • Wide inference support
  • Historical significance
tiiuae/falcon-40b-instruct

Use Cases

What TII (Falcon) is best for

Each provider has specialisms shaped by their architecture decisions, training data, and API design. Here is where TII (Falcon) performs best.

Unrestricted commercial open-source deployments
Long-context inference with linear cost
Vision-language production systems
Academic research with no licence restrictions
Foundation for domain fine-tuning

Get started

Start building with TII (Falcon)

Visit the official TII (Falcon) platform to access API keys, documentation, and pricing details.