Ling 3.0 Flash VL
Active Parameters
124B
Context Length
131K
Modality
Text
Architecture
Mixture of Experts (MoE)
License
MIT License
Release Date
4 Sept 2026
Knowledge Cutoff
-
API Pricing (per 1M)
Input: $0.00 · Output: $0.00
System Requirements
VRAM requirements for different quantization methods and context sizes
1,024 tokens
Consumer
13x RTX 4090
24GB VRAM
Datacenter
4x NVIDIA A100
80GB VRAM
Apple Silicon
3x Apple M3 Max
128GB VRAM
131,072 tokens
Consumer
18x RTX 4090
24GB VRAM
Datacenter
5x NVIDIA A100
80GB VRAM
Apple Silicon
4x Apple M3 Max
128GB VRAM
Architecture Diagram
Evaluation Benchmarks
Rank
#63
| Benchmark | Score | Rank |
|---|---|---|
Agentic Index | 0.30 | 56 |
Coding Index | 0.57 | 67 |
Intelligence Index | 0.25 | 96 |
Rankings
Overall Rank
#63
Coding Rank
#48
About Ling 3.0 Flash VL
Ling 3.0 Flash VL is a vision-language multimodal foundation model built on a Mixture-of-Experts architecture. It is optimized for efficient high-speed visual reasoning and document understanding.
Technical Specifications
Attention
Attention Structure
Multi-Layer Attention
Attention Heads
32
Key-Value Heads
32
Attention Head Dimension
128
Position Embedding
ROPE
RoPE Theta
6,000,000
Sliding Window Attention
No
Sliding Window Size
-
Sliding Window Ratio
-
Linear Attention
Yes
Linear Attention Ratio
83.3%
Normalization
RMS Normalization
Activation Function
SwigLU
Dimensions
Hidden Dimension Size
2,560
Number of Layers
42
FFN Intermediate Size (Dense)
6,144
Multi-Token Prediction Heads
-
Tokenizer
Vocabulary Size
157,184
Mixture of Experts
Total Expert Parameters
5.5B
Number of Experts
512
Active Experts
8
Shared Experts
-
FFN Intermediate Size (per Expert)
768
Dense Layers Before MoE
2