Alibaba Cloud
Qwen's international model platform and open-weight ecosystem, including the Qwen3.8 flagship line, Qwen3.8-Flash-Next and text, vision, coding, image, video, audio, embedding and rerank APIs.
Quick answer
Qwen Cloud makes Qwen available for global evaluation through an English marketplace, docs, pricing and compatible API paths.
Qwen is tracked across Qwen Studio, Qwen Cloud and official open-weight releases. Qwen Studio currently lists Qwen3.8-Max as the flagship with a 1M context and text, image and video input. Qwen's August 26 release separates Qwen3.8-Flash-Next, an open architecture preview with 125B total and 6B active parameters, from the hosted Qwen3.8-Flash production line; the release lists $0.16/M input and $0.47/M output and says API access is coming soon. Qwen3.8-2.4T-A95B and Qwen3.8-27B remain separate open-weight deployment choices with different licenses and hardware requirements.
Topic guideA source-backed map of Qwen3.8 and Qwen3.7 models, open weights, Qwen Cloud APIs, Qwen Code, Token Plan and multimodal products.
Editorial verdict
Developers evaluating the Qwen3.8 line, Qwen Cloud APIs, coding agents and multimodal Chinese model coverage from an English international platform.
Avoid assuming QwenChat, Qwen open weights and Qwen Cloud Token Plan share the same limits, data terms or commercial rights.
Qwen Cloud makes Qwen available for global evaluation through an English marketplace, docs, pricing and compatible API paths.
Free tier, pay-as-you-go API usage and Token Plan subscriptions vary by model
Free web access, Qwen Cloud billing, Token Plan, Pay-as-you-go API billing
Check the license and cloud terms for each model and deployment path.
Cloud API and self-hosted deployments have different data handling profiles.
Useful when comparing Chinese model families for coding and agent workflows.
Qwen Cloud documents OpenAI-compatible Responses, Chat Completions, Anthropic Messages and DashScope calling paths.
Use Qwen Cloud when you need one international surface for text, image, video, speech, embeddings and reranking.
Use the official research index to follow Qwen releases, open-source projects and research across coding, vision, speech, agents, safety and training methods.
Qwen's current multimodal hosted models can be evaluated for screen understanding, visual references and agent workflows; confirm the exact model and feature availability before building around it.
Qwen3.8 hosted and open-weight surfaces were rechecked on September 7, 2026. Flash-Next is an open architecture preview, while the official release describes Qwen3.8-Flash as the hosted production line with API access coming soon. Verify current pricing, infrastructure, model availability and each license separately.
DeepSeek is a reference point when the decision centers on API price and model quality.
ByteDance Ark is a cataloged comparison point for multimodal APIs, model access and cloud ecosystem fit.
Official · EN · verified 2026-08-28
Confirms the open experimental architecture, license, 125B/6B active parameters, native vision, 262K context and optional 1M extension.
Official · EN · verified 2026-09-07
Separates the open Flash-Next preview from the hosted Qwen3.8-Flash line and records the published hosted price and API-coming-soon status.
Official · EN · verified 2026-09-07
Lists Qwen3.8-Max as the flagship with 1M context and text, image and video input.
Official · EN · verified 2026-08-17
Confirms Qwen's first open Max-class model, 2.4T total and 95B active parameters, official weights and custom license.
Official · EN · verified 2026-08-17
Confirms Apache-2.0 open weights and text, image and video input support.
Official · EN · verified 2026-08-04
Confirms model ID qwen3.7-max, text-only input/output, 1M context, 65.53K maximum output, 600 RPM, 1M TPM, built-in tools, closed-source status and current 50%-off rates of $1.25/M input and $3.75/M output tokens.
Official · EN · verified 2026-06-21
Confirms model ID qwen3.7-plus, image/text/video input, text output, 1M context, 65.53K maximum output, 15K RPM, 5M TPM, built-in tools and promotional API price ranges of $0.32-$0.96/M input and $1.28-$3.84/M output tokens.
Official · EN · verified 2026-06-21
Indexes 60 official release, open-source and research entries spanning language, coding, vision, image, speech, multimodal, agent, safety and training work.
Official · EN · verified 2026-05-20
Confirms the latest preview model family update.
Official · EN · verified 2026-06-21
Confirms the current international marketplace lineup across Qwen3.7-Max/Plus, Qwen3.6, HappyHorse video, Wan and Qwen-Image image generation, CosyVoice and third-party models.
Documentation · EN · verified 2026-05-17
Confirms OpenAI SDK integration, DashScope international endpoints and model/API coverage.
Documentation · EN · verified 2026-05-17
Lists text, image/video, audio/speech, embedding and reranking model categories.
Pricing · EN · verified 2026-05-17
Confirms Token Plan subscriptions and mainstream OpenAI/Anthropic-compatible tool support.
Official · EN · verified 2026-05-20
Confirms the latest hosted proprietary Qwen family release.
Official · EN · verified 2026-05-15
Overseas support report identifies this as Qwen's international web entry.
Last checked: 2026-09-07
No approved reviews yet.