Qwen3.8-27B
- In Input modalities: TextImageVideo Out Output modalities: Text
- Reasoning· xhigh, medium, low
- Reasoning can be off
- Tool use
- Structured output
- Open weights
Prices
4 offerings · prices as listedqwen3.8-27b| Conditions/ 1M tokens | Input | Cache read | Cache write | Output |
|---|---|---|---|---|
| Standard | $0.50 | $0.10 | — | $3.00 |
| Explicit cache | — | $0.050 | $0.625 | — |
Prices from Alibaba Cloud Model Studio pricing (opens in a new tab) (checked Nov 2, 2026) · qwen3.8-27b model (opens in a new tab) (checked Nov 1, 2026)
qwen3.8-27b| Conditions/ 1M tokens | Input | Output |
|---|---|---|
| Standard | ¥3.00 | ¥12.00 |
Prices from 阿里云百炼：模型价格 (opens in a new tab) (checked Nov 2, 2026)
Standard: the price a request with no conditions pays (the headline price elsewhere on the site).
Details
- Max thinking budget
- 262.14K
- Licence
- apache-2.0
- Family
- qwen
- Model ID
Not stated on Alibaba Qwen’s pages: max input, built-in tools, knowledge cutoff, aliases.
Run locally
Weights: Qwen/Qwen3.8-27B- Parameters 27.8B
- Dense
- Native precision BF16
- Layers 64
- KV cache per token 64 KB
- Native context 256K (model card)
- Config positions 256K (config.json max_position_embeddings)
- API context 977K (maker’s API)
- Needs
- ≈ 19.2 GB (weights 17.8 + KV cache 0.5 + overhead 0.9)
- Usable
- 21.6 GB (90% of its memory)
- Generation speed
- 26–38 tok/s
- Longest context
- 46K (longer doesn’t fit)
- First token
- about 0.90–1.35 s for a 2K prompt
| Precision | Weights | Needs at 8K | On your hardware | Longest context |
|---|---|---|---|---|
| BF16 (native)maker’s file | 51.7 GB | ≈ 54.8 | ✕Doesn’t fit | — |
| Q8_0 | 27.5 GB | ≈ 29.4 | ○Needs system memory 4–6 tok/s | — |
| FP8maker’s file | 28.7 GB | ≈ 30.7 | ○Needs system memory 4–6 tok/s | — |
| Q6_K | 21.2 GB | ≈ 22.8 | ○Needs system memory 13–20 tok/s | — |
| Q5_K_M | 18.4 GB | ≈ 19.8 | ●Runs well 25–37 tok/s | 36K |
| Q5_0 | 17.8 GB | ≈ 19.2 | ●Runs well 26–38 tok/s | 46K |
| Q4_K_M | 15.7 GB | ≈ 17.0 | ●Runs well 29–43 tok/s | 81K |
| Q4_0 | 14.6 GB | ≈ 15.8 | ●Runs well 31–47 tok/s | 101K |
| IQ4_XS | 13.7 GB | ≈ 14.9 | ●Runs well 33–49 tok/s | 115K |
| Q3_K_Mquality drops | 12.6 GB | ≈ 13.8 | ●Runs well 36–54 tok/s | 133K |
| IQ3_XXSquality drops | 9.9 GB | ≈ 10.9 | ●Runs well 45–68 tok/s | 179K |
| Q2_Kquality drops | 9.7 GB | ≈ 10.7 | ●Runs well 46–69 tok/s | 182K |
| IQ2_XXSquality drops | 6.7 GB | ≈ 7.7 | ●Runs well 66–98 tok/s | 231K |
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 4090; unified memory Apple M6 (24GB or 32GB) · 32 GB.
Run commands
The maker publishes no files at this precision. Choose its native precision or one of its own builds, or quantise it yourself with llama.cpp.
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 152.7
#52 of 118 listed models · CI 150.8–154.8
- LMArena Text 1,436
#67 of 163 listed models · CI 1,431–1,441 · 27,779 votes
- EQ-Bench Creative Writing v3 1,671
#36 of 93 listed models
- EQ-Bench Longform Writing 53.3
#73 of 92 listed models
- Epoch AI, 'Capabilities & benchmarking'. Published online at epoch.ai. Retrieved from 'https://epoch.ai/benchmarks'. Licensed CC BY 4.0. · License · as of Nov 5, 2026
- LMArena (Arena), leaderboard-dataset on Hugging Face, licensed CC BY 4.0 · License · as of Oct 19 – 28, 2026
- EQ-Bench by Samuel J. Paech (eqbench.com; EQ-bench/EQ-bench-site on GitHub), MIT License. · License · as of Oct 28, 2026
- LiveBench (White et al.), release 2026-06-25, livebench.ai, licensed CC BY-SA 4.0. This board, adapted from it, is shared under CC BY-SA 4.0. · License · as of Jun 25, 2026
- changes we made
News
All news →- Aug 17, 2026Release Alibaba Qwen releases Qwen3.8-27B 1M-token context window. On Alibaba Cloud Model Studio: $0.50 input, $3.00 output per 1M tokens.
Changes
All changes →- Now priced on Cloudflare Workers AI Input $0.45 · Output $3.20
- New language model
Sources
- 阿里云百炼：模型价格in Chinese (opens in a new tab) help.aliyun.com · checked Nov 2, 2026
- Alibaba Cloud Model Studio: newly released models (opens in a new tab) alibabacloud.com · checked Nov 2, 2026
- qwen3.8-27b model (opens in a new tab) alibabacloud.com · checked Nov 1, 2026
- 硅基流动：模型价格in Chinese (opens in a new tab) siliconflow.cn · checked Oct 30, 2026
- Deep thinking - Alibaba Cloud Model Studio (opens in a new tab) docs.modelstudio.console.alibabacloud.com · checked Oct 30, 2026
- Cloudflare Workers AI pricing (opens in a new tab) developers.cloudflare.com · checked Oct 31, 2026
- Qwen3.8-27B (opens in a new tab) huggingface.co · checked Nov 2, 2026
- qwen3.8-27b Model Info-Alibaba Cloud Model Studio(Model Studio)-阿里云帮助中心in Chinese (opens in a new tab) help.aliyun.com · checked Oct 31, 2026
- Qwen/Qwen3.8-27B model card (opens in a new tab) huggingface.co · checked Nov 1, 2026
- Qwen/Qwen3.8-27B config.json (opens in a new tab) huggingface.co · checked Oct 31, 2026
- Qwen/Qwen3.8-27B files (opens in a new tab) huggingface.co · checked Oct 31, 2026
- Qwen/Qwen3.8-27B-FP8 files (opens in a new tab) huggingface.co · checked Oct 31, 2026
- Alibaba Cloud Model Studio pricing (opens in a new tab) alibabacloud.com · checked Nov 2, 2026
- Qwen/Qwen3.8-27B README.md (opens in a new tab) huggingface.co · checked Nov 4, 2026
Alibaba Qwen’s official policies: data use, privacy, terms →
Alibaba Qwen · China · qwen.ai
Related models
Snapshots, the same family and other models from this provider