Kolibri 1
Active Parameters
78B
Context Length
262K
Modality
Text
Architecture
Mixture of Experts (MoE)
License
Apache 2.0
Release Date
2 Oct 2026
Knowledge Cutoff
-
API Pricing (per 1M)
Self-hosted only
System Requirements
VRAM requirements for different quantization methods and context sizes
1,024 tokens
Consumer
8x RTX 4090
24GB VRAM
Datacenter
3x NVIDIA A100
80GB VRAM
Apple Silicon
2x Apple M3 Max
128GB VRAM
262,144 tokens
Consumer
9x RTX 4090
24GB VRAM
Datacenter
3x NVIDIA A100
80GB VRAM
Apple Silicon
2x Apple M3 Max
128GB VRAM
Architecture Diagram
Evaluation Benchmarks
No evaluation benchmarks for Kolibri 1 available.
Rankings
Overall Rank
-
Coding Rank
-
About Kolibri 1
Kolibri 1 is an open foundation language model released by Aleph Alpha featuring native causal language modeling architectures. It is designed for European multilingual processing, enterprise compliance, and general reasoning tasks.
Technical Specifications
Attention
Attention Structure
Grouped-Query Attention
Attention Heads
48
Key-Value Heads
4
Attention Head Dimension
128
Position Embedding
ROPE
RoPE Theta
10,000
Sliding Window Attention
Yes
Sliding Window Size
513
Sliding Window Ratio
80.0%
Linear Attention
No
Linear Attention Ratio
-
Normalization
RMS Normalization
Activation Function
SwigLU
Dimensions
Auxiliary Parameters
-
Hidden Dimension Size
2,560
Number of Layers
50
FFN Intermediate Size (Dense)
-
Multi-Token Prediction Heads
-
Tokenizer
Vocabulary Size
128,000
Mixture of Experts
Total Expert Parameters
3.5B
Number of Experts
384
Active Experts
6
Shared Experts
1
FFN Intermediate Size (per Expert)
512
Dense Layers Before MoE
0
Resources
About Kolibri
The Kolibri model family developed by Aleph Alpha.
Other Kolibri Models
- No related models available