Leaderboard / InclusionAI (Ling)

Ling 3.0 Tiny

Name: Ling 3.0 Tiny
Author: InclusionAI (Ling)

InclusionAI (Ling) · Ling 3.0 · released 2026-08-10 · data checked 2026-11-01 Needs verification

Ling 3.0 Tiny is a 7.9B-parameter Mixture-of-Experts model (1.4B active per token) from InclusionAI (Ling) (China), released 2026-08-10 under the MIT license. It scores 51.9 on the Small LM Index, #13 of 26 small models. It needs about 5.7 GB of memory at 4-bit with an 8K-token context.

Small LM Index

51.9

#13 of 26 scored small models

Likely 47.1–56.6 · from 3 of 6 core benchmarks

Parameters

7.9B

1.4B active · MoE

Price per 1M tokens

—

No hosted API on OpenRouter; self-host

Context window

131K

tokens

Benchmarks

Each bar shows this model's score. The tick marks the best small model on that benchmark. Sources: 1 reported by InclusionAI (Ling), 3 measured by another lab. How we label sources.

MMLU-Pro Other lab Knowledge · Measured by Edge065.8

Class best: Qwen3.5 9B 82.5

GPQA Diamond Lab-reported Knowledge73.4

Class best: Qwen3.5 9B 81.7

AIME Other lab Math · AIME 2026, measured by Edge073.3

Class best: Apriel 1.6 15B Thinker 88.0

IFBench Other lab Instruction following · Measured by Edge060.6

Class best: Granite 4.2 8B 79.3

Not reported: estimated

What the index model expects from this model's other scores. Not a measurement.

Humanity's Last Exam≈ 7

LiveCodeBench≈ 59

SWE-bench Verified≈ 34

Specifications

Lab: InclusionAI (Ling) (CN · China)
Architecture: Mixture-of-Experts
Total parameters: 7.9B
Active parameters: 1.4B
Size class: Small
Context window: 131K tokens
Input modalities: text
Reasoning: Hybrid (thinking on/off)
License: MIT
Openness: Open weights
Memory to run: ~5.7 GB at 4-bit, 8K context (more sizes)
Hugging Face: 14.3K downloads (30 days) · 566 likes

Weights on Hugging Face ↗

Compare with

Ling 3.0 Tiny vs Granite 4.2 8B (8B, index 63.5)
Ling 3.0 Tiny vs Ministral 3 8B (8B, index 55.2)
Ling 3.0 Tiny vs Edge0 8B-A1B (preview) (7.9B, index 47.2)
Ling 3.0 Tiny vs Gemma 3n E4B (7.8B, index 18.2)
Ling 3.0 Tiny vs MiMo-7B-RL (7.8B, index 40.7)
Ling 3.0 Tiny vs Gemma 4 E4B (8B, index 38.2)

Sources

Something wrong here? Report it.

Run it locally

Uses bartowski/Ling-3.0-tiny-GGUF (Q4_K_M). Swap the tag for another size: BF16, IQ2_M, IQ3_M, IQ3_XS, IQ3_XXS, IQ4_NL, IQ4_XS, Q2_K….

Ollama

ollama run hf.co/bartowski/Ling-3.0-tiny-GGUF:Q4_K_M

llama.cpp

llama-server -hf bartowski/Ling-3.0-tiny-GGUF:Q4_K_M

Then open http://localhost:8080 for a chat window.

vLLM (GPU server)

vllm serve inclusionAI/Ling-3.0-tiny

Full-precision weights, for a server GPU.

Open in LM Studio

Memory needed: weights, context cache and runtime buffers.

Quantization	4K context	32K context	131K context
16-bit F16	17 GB	17 GB	20 GB
8-bit Q8_0	9.2 GB	10.0 GB	13 GB
4-bit Q4_K_M	5.6 GB	6.4 GB	9.1 GB

Check it against your GPU, Mac or phone →