K2 Horizon MoVA 36B A4B
Active Parameters
36B
Context Length
524K
Modality
Text
Architecture
Mixture of Experts
KV Cache Rate
192 KB/tok
License
Apache 2.0
Release Date
1 Sept 2026
Knowledge Cutoff
-
API Pricing (per 1M)
Self-hosted only
Hardware Sizing
Precision
Context
Workload Memory: 22.63 GB
INT4 · 8K context
Base Weights: 18.63 GB
KV Cache: 1.50 GB
Runtime & Graphs: 2.50 GB
Minimum Required Hardware
1x NVIDIA A40 48GB
22.6 / 43.2 GB usable
2x RTX 4090 24GB
22.6 / 43.2 GB usable · TP 2
Apple M3 Pro 36GB
22.6 / 27.0 GB usable
Determine requirements for your specific workload:
Open in VRAM Calculator
Architecture Diagram
Evaluation Benchmarks
Rank
#80
| Benchmark | Score | Rank |
|---|---|---|
Agentic Index | 0.34 | 54 |
Intelligence Index | 0.25 | 107 |
Rankings
Overall Rank
#80
Coding Rank
-
About K2 Horizon MoVA 36B A4B
K2 Horizon MoVA 36B A4B is a mixture-of-experts language model activating 4B parameters per token for efficient high-capacity inference. It is optimized for complex problem-solving, context retention, and reasoning tasks.
Technical Specifications
Attention
Attention Structure
Grouped-Query Attention
Attention Heads
32
Key-Value Heads
8
Attention Head Dimension
128
Position Embedding
ROPE
RoPE Theta
10,000,000
Normalization
RMS Normalization
Activation Function
SwigLU
Dimensions
Hidden Dimension Size
2,560
Number of Layers
48
FFN Intermediate Size (Dense)
6,144
Tokenizer
Vocabulary Size
250,624
Mixture of Experts
Total Expert Parameters
4.0B
Number of Experts
100
Active Experts
8
Shared Experts
1
FFN Intermediate Size (per Expert)
768
Resources
About Kimi K2
Moonshot AI's Kimi K2 is a Mixture-of-Experts model featuring one trillion total parameters, activating 32 billion per token. Designed for agentic intelligence, it utilizes a sparse architecture with 384 experts and the MuonClip optimizer for training stability, supporting a 128K token context window.