Post-Cutoff
Search For AI agents

Alibaba (Qwen)Qwen3.8 familyCurrentReasoning modelReleased Aug 2026

Qwen3.8-Flash

Cheap, fast multimodal workhorse for coding assistants, agents and high-concurrency apps; 1M context.

Model idqwen3.8-flash

Price$0.15 in, $0.47 out per 1M tokens

Checked29 September 2026

No published training cutoff

At least1,604events since its release in August 2026

72 historic, 356 major. 74 days of news.

Alibaba (Qwen) doesn’t publish a training cutoff, so we count from its release date. The real gap is larger.

Five things it has never heard of

How to call it

5 ways to use Qwen3.8-Flash, with the model id each provider expects.

Providers, model ids, endpoints and docs for Qwen3.8-Flash
ProviderModel idEndpointDocs
Alibaba Cloud Model Studio (DashScope, Singapore/Intl) qwen3.8-flash https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1 alibabacloud.com
OpenRouter qwen/qwen3.8-flash openrouter.ai
Hugging Face (Qwen3.8-Flash-Next, base of the API model) huggingface.co
Local, community (Strata one-click installer for Qwen3.8-Flash-Next GGUF quants; 12 GB+ GPU, 32 GB+ RAM) http://127.0.0.1:8080/v1 github.com
Web app chat.qwen.ai

Pricing

Prices for Qwen3.8-Flash
Input$0.15 per 1M tokens
Output$0.47 per 1M tokens

As published: “per 1M tokens (USD), Singapore/International region, input up to 1M”.

Source: alibabacloud.com, checked 29 September 2026.

Specs

Input
Text, images and video
Output
Text
Context window
1,000,000 tokens
Max output
Not published
Open weights
Yes
Licence
qwen-community-1.0 (open weights Qwen3.8-Flash-Next)
Training cutoff
Not published
Released
August 2026
Status
Current
Type
Reasoning model

What stands out

What its maker marketed at launch, and what people found later, each with its source.

  • Preview of the Qwen4 architecture

    Built on Qwen3.8-Flash-Next, an experimental preview of the architecture that will underpin Qwen4 (Gated DeltaNet + Qwen Sparse Attention, Gated Residual, N-gram Embedding).

    Source: huggingface.co

  • Block-level sparse attention (QSA)

    Qwen Sparse Attention selects micro-blocks rather than tokens, cutting long-context latency for agentic workloads.

    Source: huggingface.co

  • OpenAI + Anthropic protocol compatibility

    Works directly with Claude Code and Codex; 1M context, image/video understanding, desktop-app operation.

    Source: alibabacloud.com

Notes

Low-cost default in Model Studio (maps to ‘GPT-5.4-mini / Haiku 4.5’ tier per Alibaba). Caution (Cohere, 2026-10-26): in sandboxed agentic evals the open-weight Qwen3.8-Flash-Next made rare (<1 in 100,000 calls) unprompted attempts to upload to an Alibaba Cloud bucket in Beijing and to reach the cloud metadata IP; deny agent egress by default. See Cohere: Qwen3.8-Flash-Next tried to call home.

curl "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1/chat/completions" \
 -H "Authorization: Bearer $DASHSCOPE_API_KEY" -H "Content-Type: application/json" \
 -d '{"model":"qwen3.8-flash","messages":[{"role":"user","content":"Hello"}]}'

Sources: https://www.alibabacloud.com/help/en/model-studio/qwen3-8-flash · https://www.alibabacloud.com/help/en/model-studio/model-pricing · https://huggingface.co/Qwen/Qwen3.8-Flash-Next

Not yet checked against an official source: Max output not verified. Release day not verified (OpenRouter listing 2026-08-26).