Leaderboard / Alibaba (Qwen)

Qwen3 8B

Name: Qwen3 8B
Author: Alibaba (Qwen)

Alibaba (Qwen) · Qwen3 · released 2025-04-29 · data checked 2026-11-01 Needs verification

Qwen3 8B is a 8.2B-parameter dense model from Alibaba (Qwen) (China), released 2025-04-29 under the Apache-2.0 license. It scores 48.5 on the Small LM Index, #14 of 26 small models. It needs about 6.8 GB of memory at 4-bit with an 8K-token context.

Small LM Index

48.5

#14 of 26 scored small models

Likely 43.6–53.2 · from 3 of 6 core benchmarks

Parameters

8.2B

Dense

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: 3 reported by Alibaba (Qwen). How we label sources.

GPQA Diamond Lab-reported Knowledge62.0

Class best: Qwen3.5 9B 81.7

AIME Lab-reported Math · AIME 2025 (thinking)67.3

Class best: Apriel 1.6 15B Thinker 88.0

LiveCodeBench Lab-reported Coding · LiveCodeBench v557.5

Class best: Granite 4.2 8B 73.2

Not reported: estimated

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

MMLU-Pro≈ 71

Humanity's Last Exam≈ 6

SWE-bench Verified≈ 29

Specifications

Lab: Alibaba (Qwen) (CN · China)
Architecture: Dense transformer
Total parameters: 8.2B
Active parameters: All (dense)
Size class: Small
Context window: 131K tokens
Input modalities: text
Reasoning: Hybrid (thinking on/off)
License: Apache-2.0
Openness: Open weights
Memory to run: ~6.8 GB at 4-bit, 8K context (more sizes)
Hugging Face: 7.5M downloads (30 days) · 2.2K likes

Weights on Hugging Face ↗

Compare with

Qwen3 8B vs DeepSeek-R1-0528-Qwen3-8B (8.2B, index 53.1)
Qwen3 8B vs Granite 4.2 8B (8B, index 63.5)
Qwen3 8B vs Ministral 3 8B (8B, index 55.2)
Qwen3 8B vs Gemma 4 E4B (8B, index 38.2)

Sources

https://arxiv.org/abs/2505.09388

Something wrong here? Report it.

Run it locally

Uses Qwen/Qwen3-8B-GGUF (Q4_K_M). Swap the tag for another size: Q4_K_M, Q5_0, Q5_K_M, Q6_K, Q8_0.

Ollama

ollama run hf.co/Qwen/Qwen3-8B-GGUF:Q4_K_M

llama.cpp

llama-server -hf Qwen/Qwen3-8B-GGUF:Q4_K_M

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

MLX (Mac)

pip install -U mlx-lm
mlx_lm.chat --model mlx-community/Qwen3-8B-4bit

vLLM (GPU server)

vllm serve Qwen/Qwen3-8B

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	18 GB	22 GB	36 GB
8-bit Q8_0	10.0 GB	14 GB	29 GB
4-bit Q4_K_M	6.2 GB	10 GB	25 GB

Check it against your GPU, Mac or phone →