--- title: "Qwen 3.8-Max: Release Date, Specs, and How to Access It (2026)" slug: qwen-3-8-max-release-date-specs-how-to-access-2026 description: "Alibaba unveiled Qwen 3.8-Max at WAIC: 2.4T parameters, multimodal, claimed second only to Claude Fable 5. What's confirmed, what isn't, and how to try it." author: "Yotta Labs" date: 2026-07-20 categories: ["Inference"] canonical: https://www.yottalabs.ai/post/qwen-3-8-max-release-date-specs-how-to-access-2026 --- # Qwen 3.8-Max: Release Date, Specs, and How to Access It (2026) ![](https://cdn.sanity.io/images/wy75wyma/production/e771cb4ef9f67e8a4042542c21906bb284677b67-1200x627.png) Alibaba's Qwen Team previewed Qwen 3.8-Max on July 19, 2026, and it is the biggest model announcement of the summer: 2.4 trillion parameters, multimodal, and a claim that it trails only Anthropic's Claude Fable 5. Here is what is actually confirmed, what is still a vendor claim, and how to start evaluating it today. ## TL;DR - Announced: July 19, 2026, at the World AI Conference in Shanghai. Preview access, not general availability - Size: 2.4 trillion parameters, sparse Mixture-of-Experts. Active parameters per token not disclosed - Multimodal: text plus visual inputs confirmed. Coverage differs on the full list (video, documents, speech, image generation have all been reported); Alibaba has not published a spec sheet - Context window: not yet confirmed - Performance claim: "second only to Fable 5," Alibaba's own words. No benchmark table published yet - Open source: not yet. Qwen's official announcement says it is "going open-weight soon," with no date or license named - Access today: preview through Alibaba's Token Plan subscription and the Qoder and QoderWork platforms at a reported 10% of standard pricing. No standalone API pricing announced - API compatibility: OpenAI spec and Anthropic spec, same as the rest of the Max line - Running Qwen in production now: Qwen 3.7-Max is live on [Yotta AI Gateway](https://www.yottalabs.ai/ai-gateway) ## What Qwen 3.8-Max Is Qwen 3.8-Max is the next flagship in Alibaba's Qwen line, previewed two months after Qwen 3.7-Max shipped. Where 3.7-Max was framed entirely around long-horizon agent workloads, 3.8-Max adds a second headline: native multimodality. Text plus visual inputs is confirmed; early coverage also reports video, document, speech, and image generation support, but Alibaba has not published a spec sheet, so treat the full modality list as unsettled. The stated target workloads are coding, full-stack development, data analysis, and office workflows. That is a direct continuation of the agent positioning the 3.7 line established, now with visual inputs in scope. The scale claim matters for context: at 2.4 trillion parameters, Qwen 3.8-Max is the second-largest publicly known model, behind Moonshot's Kimi K3 at 2.8 trillion, which launched as an open-weight release the same week. The timing is not a coincidence. The frontier race among Chinese labs is compressing release cycles. One number Alibaba did not publish: active parameters per token. For a sparse MoE model, that is the figure that determines serving cost and latency. Until it is disclosed, there is no way to estimate what 3.8-Max costs to run at scale. ## Qwen 3.8 Release Date Alibaba previewed Qwen 3.8-Max on July 19, 2026, at the World AI Conference (WAIC) in Shanghai. Some outlets date the announcement July 20 due to timezone differences. This is a preview, not general availability. Standard API access and pricing have not been announced. Source: [Qwen's official announcement](https://x.com/Alibaba_Qwen/status/2078759124914098291) If you are seeing "Qwen 3.8" and "Qwen 3.8-Max" used interchangeably in coverage, the model shown at WAIC is Qwen 3.8-Max, the flagship tier. ## Is Qwen 3.8 Open Source? Not yet, and this is the most interesting open question of the release. Every Max-tier Qwen model so far has stayed closed. [Qwen 3.7-Max is API-only](https://www.yottalabs.ai/post/qwen-3-7-max-release-date-features-open-source-status-and-how-to-access-2026), and the open-weight line continued separately with Qwen 3.6. But this time the promise is explicit and public: Qwen's own announcement says Qwen 3.8 is "launching and going open-weight soon." No date, no license, no repository yet. If that happens, it breaks the pattern that has defined the Qwen lineup, and it would be a direct answer to Kimi K3 shipping 2.8 trillion open-weight parameters the same week. Until there is a Hugging Face repository and a license, treat open-weight 3.8 as a promise, not a plan. We will update this post when it lands either way. If you need open weights today, the practical options remain Qwen 3.6, [GLM 5.2](https://www.yottalabs.ai/post/glm-5-2-vs-qwen-3-7-max-open-weights-vs-proprietary-2026), and now Kimi K3. ## Qwen 3.8-Max Benchmarks There are none yet, and that is worth saying plainly. Alibaba's exact words: Qwen 3.8 is "one of the most powerful models available today, comparable to leading frontier AI models, second only to Fable 5," referring to Anthropic's Claude Fable 5. That ranking rests on internal evaluations. No benchmark table has been published, no model card, no third-party test. Every performance figure in circulation right now is a vendor claim. For calibration, the verified scores of its predecessor: Qwen 3.7-Max posted 92.4 on GPQA Diamond, 80.4 on SWE-bench Verified, and 69.7 on Terminal-Bench 2.0, competitive with Claude Opus 4.6 on most of the agentic suite. If 3.8-Max improves on that baseline while adding multimodality, the ranking claim is plausible. Plausible is not verified. The right move for production teams: wait for the benchmark table, then run your own workload against it. Vendor rankings, including this one, are directional at best. ## How to Access Qwen 3.8-Max Right now, access is preview-only through Alibaba's own channels: the Token Plan subscription and the Qoder and QoderWork platforms, at a reported 10% of standard pricing during the preview window. There is no standalone API pricing yet. The model speaks both the OpenAI API spec and the Anthropic API spec, consistent with the rest of the Max line, so existing client code should port with minimal changes once broader API access opens. For teams that want Qwen-class agent capability in production today, [Qwen 3.7-Max is live on Yotta AI Gateway](https://www.yottalabs.ai/ai-gateway): one API key, OpenAI-compatible and Anthropic-compatible endpoints, routed alongside Claude, DeepSeek, GLM, and the rest of the catalog. We broke down the full deployment decision in [how to run Qwen 3.7 in production](https://www.yottalabs.ai/post/how-to-run-qwen-3-7-in-production), and the same logic will apply to 3.8 as access expands. ## Qwen 3.8-Max vs Qwen 3.7-Max What actually changed, based on what has been disclosed: The practical read: 3.7-Max is the Qwen you can build on today. 3.8-Max is the one to evaluate the moment real benchmarks and API pricing appear. If cost is the deciding factor between frontier models, our [Qwen 3.7 Max vs Claude Opus 4.6 pricing breakdown](https://www.yottalabs.ai/post/qwen-3-7-max-vs-claude-opus-4-6-pricing-benchmarks-2026) shows how the current generation compares. ## Qwen 3.8-Max vs Kimi K3 The two biggest model announcements of July landed days apart, and they represent opposite bets. Moonshot's Kimi K3 shipped first as a 2.8 trillion parameter open-weight release, currently the largest publicly known open model: download it, self-host it, fine-tune it. Qwen 3.8-Max previewed days later at 2.4 trillion parameters as a closed, managed service with an open-weight version promised to follow. For production teams, the choice is the same one the [GLM 5.2 vs Qwen 3.7-Max](https://www.yottalabs.ai/post/glm-5-2-vs-qwen-3-7-max-open-weights-vs-proprietary-2026) matchup posed: open weights give you deployment control, fine-tuning rights, and GPU-level cost management; a closed frontier API gives you the vendor's best model with zero infrastructure lift. Neither model has third-party benchmarks yet, so any head-to-head performance ranking between them is premature. We will publish a full comparison once both have verifiable numbers. ## Frequently Asked Questions **When did Qwen 3.8 release?** Alibaba previewed Qwen 3.8-Max on July 19, 2026, at WAIC in Shanghai. It is in preview, not general availability. **Is Qwen 3.8 open source?** Not currently. Alibaba has promised an open-weight release "soon" but has not named a date or license. Its predecessor, Qwen 3.7-Max, remains closed and API-only. **How big is Qwen 3.8-Max?** 2.4 trillion parameters in a sparse Mixture-of-Experts architecture. Active parameters per token, the number that drives serving cost, has not been disclosed. **Is Qwen 3.8-Max better than Qwen 3.7-Max?** Alibaba says it is, claiming it trails only Claude Fable 5 overall. No benchmark table has been published, so the claim cannot be verified yet. Qwen 3.7-Max's published scores remain the verified baseline for the family. **Can I use Qwen 3.8-Max through an API?** Preview access runs through Alibaba's Token Plan and Qoder platforms. Standard API access and pricing have not been announced. The model supports both OpenAI and Anthropic API specs. **Is Qwen 3.8 bigger than Kimi K3?** No. Kimi K3 is 2.8 trillion parameters and open-weight; Qwen 3.8-Max is 2.4 trillion and currently closed preview. Total parameter count says little about quality or serving cost on its own, especially for sparse MoE models where active parameters matter more. **How can I run Qwen models in production today?** Qwen 3.7-Max and Qwen 3.6 Plus are both live on [Yotta AI Gateway](https://www.yottalabs.ai/ai-gateway) with OpenAI-compatible and Anthropic-compatible endpoints, and open-weight Qwen 3.6 can be self-hosted on [Yotta GPU Pods](https://www.yottalabs.ai/pricing). ## Bottom Line Qwen 3.8-Max is a serious announcement wrapped in unverified claims. The scale is real, the multimodal direction is real, and the preview is live. The "second only to Claude Fable 5" ranking, the open-weight promise, and the production economics are all still open questions. Watch for three things: a published benchmark table, standalone API pricing, and an actual open-weight release. Any one of those turns this from news into a deployment decision. Until then, the Qwen you can ship on is 3.7-Max, [live on Yotta AI Gateway today](https://www.yottalabs.ai/ai-gateway). Start there, and this post will be updated as 3.8 access expands.