The first lineup of budget-concious, state-of-the-art open models from Aqui, now with an MIT license.
AI & ML interests
Fast, cheap, AWESOME AI platforms and models
Recent Activity
Organization Card
Welcome!
- Aqui-open0-3: Our flagship open model series, including our first open MoE model. State-of-the-art performance while using 2x less VRAM. Find more about it here
- Aqui-open0-1 (1.5B, 7.5B): The best small open models, trained from scratch. With quantization, the 1.5B can run on a smartphone smoothly. Find more about them here
- Aqui-v4.0 series: Our lineup of flagship proprietary models, pushing the frontier with techniques like AutoSummon. Find more about them here [PT-BR]
models
7
aquigpt/open0-2.5
Text Generation
•
33B
•
Updated
•
8
•
1
aquigpt/open1-7.5B-Instruct
Text Generation
•
7B
•
Updated
•
3
•
3
aquigpt/open1-1.5B-Instruct
Text Generation
•
1B
•
Updated
•
2
•
3
aquigpt/open0-2-lite
Text Generation
•
2B
•
Updated
•
18
•
3
aquigpt/open0-2
Text Generation
•
21B
•
Updated
•
4
•
2
aquigpt/aqui-vl-12b
12B
•
Updated
•
1
•
1
aquigpt/aqui-vl-24b
Updated
•
1
datasets
0
None public yet