Qwen3-8B
- In Input modalities: Text Out Output modalities: Text
- Reasoning
- Reasoning can be off
- Tool use
- Structured output
- Open weights
Prices
3 offerings · prices as listed| Conditions/ 1M tokens | Input | Output |
|---|---|---|
| Standard no longer listed since Oct 11, 2026 | $0.18 | $0.70 |
| Reasoning off no longer listed since Oct 11, 2026 | — | $0.70 |
| Reasoning on no longer listed since Oct 11, 2026 | — | $2.10 |
Prices from qwen3-8b model (opens in a new tab) (checked Oct 5, 2026) · Alibaba Cloud Model Studio pricing (opens in a new tab) (checked Oct 11, 2026)
| Conditions/ 1M tokens | Input | Output |
|---|---|---|
| Standard no longer listed since Oct 11, 2026 | $0.072 | $0.287 |
| Reasoning off no longer listed since Oct 11, 2026 | — | $0.287 |
| Reasoning on no longer listed since Oct 11, 2026 | — | $0.717 |
Prices from Alibaba Cloud Model Studio pricing (opens in a new tab) (checked Oct 11, 2026) · qwen3-8b model (opens in a new tab) (checked Oct 5, 2026)
| Conditions/ 1M tokens | Input | Output |
|---|---|---|
| Standard no longer listed since Oct 11, 2026 | ¥0.50 | — |
| Reasoning off no longer listed since Oct 11, 2026 | — | ¥2.00 |
| Reasoning on no longer listed since Oct 11, 2026 | — | ¥5.00 |
Prices from 阿里云百炼：模型价格 (opens in a new tab) (checked Oct 6, 2026)
Standard: the price a request with no conditions pays (the headline price elsewhere on the site).
Details
- Knowledge cutoff
- Apr 2025
- Licence
- apache-2.0
- Family
- qwen
- Model ID
Not stated on Alibaba Qwen’s pages: max input, reasoning efforts, max thinking budget, built-in tools, aliases.
Run locally
Weights: Qwen/Qwen3-8B- Parameters 8.2B
- Dense
- Native precision BF16
- Layers 36
- KV cache per token 144 KB
- Native context 32K (model card)
- With YaRN 128K (model card)
- Config positions 40K (config.json max_position_embeddings)
- API context 128K (maker’s API)
- Needs
- ≈ 17.1 GB (weights 15.3 (maker’s file) + KV cache 1.1 + overhead 0.8)
- Usable
- 21.6 GB (90% of its memory)
- Generation speed
- 29–43 tok/s
- Longest context
- 39K (longer doesn’t fit)
- First token
- about 0.26–0.40 s for a 2K prompt
| Precision | Weights | Needs at 8K | On your hardware | Longest context |
|---|---|---|---|---|
| BF16 (native)maker’s file | 15.3 GB | ≈ 17.1 | ●Runs well 29–43 tok/s | 39K |
| Q8_0maker’s file | 8.1 GB | ≈ 9.7 | ●Runs well 51–76 tok/s | 128K |
| FP8maker’s file | 8.8 GB | ≈ 10.4 | ●Runs well 47–71 tok/s | 128K |
| Q6_Kmaker’s file | 6.3 GB | ≈ 7.9 | ●Runs well 64–95 tok/s | 128K |
| Q5_K_Mmaker’s file | 5.4 GB | ≈ 7.1 | ●Runs well 71–107 tok/s | 128K |
| Q5_0maker’s file | 5.3 GB | ≈ 7.0 | ●Runs well 73–109 tok/s | 128K |
| Q4_K_Mmaker’s file | 4.7 GB | ≈ 6.3 | ●Runs well 81–121 tok/s | 128K |
| Q4_0 | 4.3 GB | ≈ 5.9 | ●Runs well 87–130 tok/s | 128K |
| INT4 (AWQ / GPTQ)maker’s file | 5.7 GB | ≈ 7.3 | ●Runs well 69–103 tok/s | 128K |
| IQ4_XS | 4.1 GB | ≈ 5.7 | ●Runs well 91–136 tok/s | 128K |
| Q3_K_Mquality drops | 3.7 GB | ≈ 5.4 | ●Runs well 97–145 tok/s | 128K |
| IQ3_XXSquality drops | 2.9 GB | ≈ 4.5 | ●Runs well 116–174 tok/s | 128K |
| Q2_Kquality drops | 2.9 GB | ≈ 4.5 | ●Runs well 118–177 tok/s | 128K |
| IQ2_XXSquality drops | 2.0 GB | ≈ 3.6 | ●Runs well 152–228 tok/s | 128K |
Click a row to choose a precision. “Maker’s file” sizes are the maker’s own files; others are parameters × bits per weight.
Compared with other hardware
Each at the highest precision it runs. Click one for details.
Smallest that runs it (Q4_K_M, 8K): GPU GeForce RTX 3080; unified memory Apple M5 · 16 GB.
Run commands
vllm serve Qwen/Qwen3-8B --max-model-len 8192
Memory to fine-tune
Batch 1, 2,048-token sequences, gradient checkpointing; LoRA and QLoRA train rank-16 adapters on the attention projections.
Estimates; facts from the maker’s Hugging Face repositories, as of 2026-11-06. How it’s estimated
Cited scores
From third-party leaderboards, not measured by us- Epoch Capabilities Index 132.2
#112 of 118 listed models · CI 129.1–135.1
- Epoch AI · GPQA Diamond 56.8%
#113 of 168 listed models · ±2.93
Also listed: reasoning off 47.5%
- Epoch AI · OTIS Mock AIME 2024–2025 56.1%
#106 of 159 listed models · ±6.45
Also listed: reasoning off 22.2%
- Epoch AI, 'Capabilities & benchmarking'. Published online at epoch.ai. Retrieved from 'https://epoch.ai/benchmarks'. Licensed CC BY 4.0. · License · as of Nov 4 – 5, 2026
- changes we made
Changes
All changes →- Context window 256K → 131.07K
Sources
- Alibaba Cloud Model Studio pricing (opens in a new tab) alibabacloud.com · checked Oct 11, 2026
- 阿里云百炼：模型价格in Chinese (opens in a new tab) help.aliyun.com · checked Oct 6, 2026
- Text generation models - Qwen, DeepSeek, GLM | Model Studio - Alibaba Cloud Model Studio - Alibaba Cloud Documentation Center (opens in a new tab) alibabacloud.com · checked Oct 10, 2026
- qwen3-8b model (opens in a new tab) alibabacloud.com · checked Oct 5, 2026
- models.dev: qwen3-8b (community data, MIT) (opens in a new tab) github.com · checked Sep 30, 2026
- qwen3-8b - Alibaba Cloud Model Studio (opens in a new tab) docs.modelstudio.console.alibabacloud.com · checked Oct 9, 2026
- Qwen/Qwen3-8B model card (opens in a new tab) huggingface.co · checked Oct 31, 2026
- Qwen/Qwen3-8B config.json (opens in a new tab) huggingface.co · checked Nov 2, 2026
- Qwen/Qwen3-8B files (opens in a new tab) huggingface.co · checked Oct 31, 2026
- Qwen/Qwen3-8B-AWQ files (opens in a new tab) huggingface.co · checked Nov 2, 2026
- Qwen/Qwen3-8B-FP8 files (opens in a new tab) huggingface.co · checked Oct 31, 2026
- Qwen/Qwen3-8B-GGUF files (opens in a new tab) huggingface.co · checked Nov 2, 2026
- Qwen/Qwen3-8B-MLX-4bit files (opens in a new tab) huggingface.co · checked Nov 3, 2026
- Qwen/Qwen3-8B-MLX-6bit files (opens in a new tab) huggingface.co · checked Oct 31, 2026
- Qwen/Qwen3-8B-MLX-8bit files (opens in a new tab) huggingface.co · checked Oct 31, 2026
- Qwen/Qwen3-8B-MLX-bf16 files (opens in a new tab) huggingface.co · checked Oct 31, 2026
- Qwen/Qwen3-8B model card (opens in a new tab) huggingface.co · checked Nov 3, 2026
- 模型上下架与更新-大模型服务平台百炼-阿里云_大模型服务平台百炼(Model Studio)-大模型服务平台百炼(Model Studio)-阿里云帮助中心in Chinese (opens in a new tab) help.aliyun.com · checked Oct 30, 2026
- Qwen/Qwen3-8B README.md (opens in a new tab) huggingface.co · checked Nov 4, 2026
Alibaba Qwen’s official policies: data use, privacy, terms →
Some values on this page are still waiting to be checked against official pages: a filled dot is verified, a hollow one is not yet (with or without a source on file). How we verify data
Alibaba Qwen · China · qwen.ai
Related models
Snapshots, the same family and other models from this provider