# Qwen3.8-27B: Self-Hosting Hardware, Scenarios and Commercial License

> A comparatively deployable native vision-language model for private coding, computer-use, document and general agent workloads.

- Verified: 2026-09-12
- Released: 2026-08-14
- Canonical: https://chinaapi.ai/open-model-deployment/qwen3.8-27b/
- Back to directory: https://chinaapi.ai/open-model-deployment/
- Machine-readable dataset: https://chinaapi.ai/data/open-model-deployment.json

## Model facts

- Vendor: Alibaba Qwen
- Parameters: 27B total; 27B dense active
- Native precision: BF16 plus an official FP8 checkpoint; the two repositories are precision variants of one model specification
- Context: 256K tokens
- Category: General and agentic foundation models
- Modalities: text, image, video → text
- Serving paths: Transformers, vLLM, SGLang
- Official weights: BF16 / original: https://huggingface.co/Qwen/Qwen3.8-27B?utm_source=chinaapi&utm_medium=research&utm_campaign=open-model-deployment; Official FP8: https://huggingface.co/Qwen/Qwen3.8-27B-FP8?utm_source=chinaapi&utm_medium=research&utm_campaign=open-model-deployment
- Model documentation: https://huggingface.co/Qwen/Qwen3.8-27B?utm_source=chinaapi&utm_medium=research&utm_campaign=open-model-deployment

## Deployment evidence

- **Official launch / minimum — evidence E:** BF16 weights alone are about 54GB before runtime and KV cache. The vendor publishes BF16 and FP8 checkpoints but no absolute minimum GPU configuration; community quantizations are not treated as an official minimum.
- **200-person team — evidence E:** Start capacity testing with two redundant 48–80GB-class workers or an equivalent tensor-parallel design, then size from the real vision, context and output mix; this is a planning estimate, not an official recommendation.
- **Commercial API — evidence C:** Official vLLM and SGLang launch paths are published. A public service still needs redundant replicas, admission control and separate short-context, vision, 262K-context and tool-call benchmarks. Source: https://huggingface.co/Qwen/Qwen3.8-27B?utm_source=chinaapi&utm_medium=research&utm_campaign=open-model-deployment

## Scenario evidence

- **Coding and repository agents — Public evidence:** The official model card reports Terminal Bench 2.1, SWE-bench Pro, DeepSWE, NL2Repo and QwenSWEBench results; these are vendor-reported, not ChinaAPI reproductions. Source: https://huggingface.co/Qwen/Qwen3.8-27B?utm_source=chinaapi&utm_medium=research&utm_campaign=open-model-deployment
- **Computer-use and mobile agents — Public evidence:** The official card publishes OSWorld, WebArena and AndroidWorld evaluations for agent interaction with visual interfaces. Source: https://huggingface.co/Qwen/Qwen3.8-27B?utm_source=chinaapi&utm_medium=research&utm_campaign=open-model-deployment
- **Private multimodal assistant — ChinaAPI inference:** The dense 27B shape and native image/video inputs make it the more practical of the two Qwen3.8 open releases for controlled enterprise deployment.

## Commercial-use check

- License: Apache-2.0 — https://huggingface.co/Qwen/Qwen3.8-27B/blob/main/LICENSE?utm_source=chinaapi&utm_medium=research&utm_campaign=open-model-deployment
- MaaS / hosted service: No model-specific MaaS restriction identified in Apache-2.0.
- Attribution: Provide the license and required notices, preserve attribution notices, and mark modified files; trademark rights are not granted.

## Avoid or validate first

- Assuming a 27B weight-memory estimate includes KV cache and vision processing
- Promising full 262K context or production concurrency from a single workstation without load tests

## Known limitations and open questions

- No ChinaAPI hardware reproduction
- The official FP8 repository is a precision variant, not a third Qwen3.8 model size
- Native 262K and extended 1M context materially increase memory demand

## Evidence boundary

Loading weights, completing a first forward pass and meeting a production latency/SLA are separate thresholds. The 200-person and commercial tiers require workload-specific measurement. License summaries are product research, not legal advice.

Generated by GENERATED BY scripts/gen_open_model_deployment.py.