Leaderboard / InclusionAI (Ling)
LLaDA2.2 mini
InclusionAI (Ling) · LLaDA2 · released 2026-09-05 · data checked 2026-11-01
LLaDA2.2 mini is a 16.3B-parameter Mixture-of-Experts model (1.4B active per token) from InclusionAI (Ling) (China), released 2026-09-05 under the Apache-2.0 license. It scores 29.7 on the Small LM Index, #33 of 34 medium models. It needs about 11 GB of memory at 4-bit with an 8K-token context.
Diffusion language model (MoE).
Benchmarks
Each bar shows this model's score. The tick marks the best medium model on that benchmark. Sources: 6 reported by InclusionAI (Ling). How we label sources.
Not reported: estimated
What the index model expects from this model's other scores. Not a measurement.
Specifications
- Lab
- InclusionAI (Ling) (CN · China)
- Architecture
- Mixture-of-Experts
- Total parameters
- 16.3B
- Active parameters
- 1.4B
- Size class
- Medium
- Context window
- 131K tokens
- Input modalities
- text
- Reasoning
- Instruct (no thinking)
- License
- Apache-2.0
- Openness
- Open weights
- Memory to run
- ~11 GB at 4-bit, 8K context (more sizes)
- Hugging Face
- 4K downloads (30 days) · 39 likes
Compare with
- LLaDA2.2 mini vs Apriel 1.6 15B Thinker (14.9B, index 60.9)
- LLaDA2.2 mini vs Phi-4-reasoning-plus (14.7B, index 54.4)
- LLaDA2.2 mini vs Phi-4 (14.7B, index 44.7)
- LLaDA2.2 mini vs Ministral 3 14B (14B, index 60.0)
Sources
Something wrong here? Report it.
Run it locally
No GGUF build from a well-known publisher yet.
vllm serve inclusionAI/LLaDA2.2-miniMemory needed: weights, context cache and runtime buffers.
| Quantization | 4K context | 32K context | 131K context |
|---|---|---|---|
| 16-bit F16 | 34 GB | 35 GB | 39 GB |
| 8-bit Q8_0 | 18 GB | 19 GB | 24 GB |
| 4-bit Q4_K_M | 11 GB | 12 GB | 16 GB |