Qwen 3.8 API Access: Pricing, Token Plan & Options (2026)

For teams trying to get Qwen 3.8 into an actual application, not just read about benchmarks.

Short version: there is no standalone Qwen 3.8 API yet. If you came here looking for an endpoint and a per-token price, neither exists as of early August 2026. What does exist is a preview program with unusually cheap access, a promise of open weights with no date attached, and a few sensible ways to build with the Qwen family today without betting your roadmap on a preview.

Here's the full picture, including what we'd actually do in production right now.

TL;DR

Where Qwen 3.8 actually stands

Alibaba announced Qwen 3.8-Max on July 19, 2026 at WAIC in Shanghai: a 2.4 trillion parameter sparse MoE model, multimodal, with active parameter count undisclosed. The launch came with a bold internal claim ("second only to Fable 5") and no published benchmark table. We broke down what's verified and what isn't in our Qwen 3.8 benchmarks post.

The part that matters for builders: this is a preview launch. There is no general availability API, no standard per-token rate card, and no way to sign up for ordinary metered access the way you would with Qwen 3.7-Max or any other production model.

The three ways to touch Qwen 3.8 today

1. Alibaba's Token Plan. A credit-based monthly subscription, not a metered API. During the preview window, Qwen 3.8-Max reportedly runs at 10% of standard credit consumption. Third-party reporting puts the Personal Edition tiers at roughly $6, $20, and $70 per month (39, 139, and 499 CNY), with an additional 80% off credit consumption between 22:00 and 08:00 UTC+8 (eesel's pricing breakdown has the details). Fine for evaluation. Not something you can build a production SLA on, and Alibaba has been explicit that preview pricing is promotional.

2. Qoder and QoderWork. Alibaba's coding products carry the preview too. If your use case is agentic coding, this is the fastest way to feel the model out.

3. Wait for the API or the weights. The Max line ships with OpenAI-compatible and Anthropic-compatible API specs, so when the GA endpoint lands, switching a test harness over should be trivial. Alibaba has also said Qwen 3.8 is "going open-weight soon," with no date and no license named. Until weights land on Hugging Face, no third-party inference provider can serve it, which is why you won't find Qwen 3.8 on any gateway or serverless platform yet, ours included.

What to run in production while you wait

Two realistic paths, depending on how you like to operate.

Managed API today: Qwen 3.7-Max. Still the production flagship of the family, with published benchmarks (92.4 GPQA Diamond, 80.4 SWE-bench Verified) and real per-token pricing at $1.25 per million input tokens and $3.75 per million output. It's live on the Yotta AI Gateway alongside Claude, DeepSeek, GLM, and the rest of the Qwen line, behind one OpenAI-compatible key. When Qwen 3.8 gets a GA API, a gateway setup means swapping a model string, not rewriting a client. We compared costs in detail in Qwen 3.7-Max vs Claude Opus 4.6.

Self-hosted today: open-weight Qwen. If your requirement is weights you control, the current open Qwen releases already run well on rented GPUs. Our guide to running Qwen 3.7 in production covers engine choice and GPU sizing, and the same playbook will apply to Qwen 3.8 the day weights actually ship.

The wrong move is architecting around a preview. Credit-subscription access with promotional pricing is for evaluation. Point your evals at it, keep production on something with a rate card.

What to watch next

Three events will change this page, and we'll update it when they happen. A GA API with published per-token pricing (the preview's "10% of standard" phrasing implies a standard rate exists internally). An open-weight release on Hugging Face, which opens the door to third-party providers and self-hosting overnight. And independent benchmark coverage, since right now the only numbers are Alibaba's own internal evals.

FAQ

Is there a Qwen 3.8 API? Not a standalone one. As of August 2, 2026, access is preview-only through Alibaba's Token Plan subscription and the Qoder/QoderWork products. No GA endpoint, no per-token rate card.

How much does Qwen 3.8 cost? There's no published per-token price. Preview access reportedly runs at 10% of standard credit rates inside Token Plan subscriptions that start around $6/month. Treat those numbers as promotional and temporary.

Is Qwen 3.8 open source? Not yet. Alibaba says the model is going open-weight, but hasn't named a date or a license. Qwen 3.8-Max itself is closed, same as 3.7-Max before it.

Can I download Qwen 3.8 from Hugging Face? No. Nothing has been published there yet. When weights land, expect the ecosystem (vLLM, SGLang, third-party APIs) to move fast, since that's exactly what happened with prior open Qwen releases.

What is Alibaba's Token Plan? A monthly credit subscription for Qwen model access, sold in Lite, Standard, and Pro tiers. It's how Alibaba is gating Qwen 3.8-Max preview access, at reported 10% credit consumption plus a night discount in the UTC+8 window.

Can I get Qwen 3.8 through a gateway or serverless provider? Not yet, anywhere. With no GA API and no open weights, third-party providers have nothing to serve. Qwen 3.7-Max, Qwen3-Max, and the open Qwen line are available through the Yotta AI Gateway today.

Is Qwen 3.8 better than Kimi K3? Early independent signals are mixed, and official benchmarks don't exist yet. We ran the comparison in Qwen 3.8 vs Kimi K3.

What should I build on today if I want Qwen 3.8 later? Anything OpenAI-compatible. The Max line ships OpenAI-spec and Anthropic-spec APIs, so a gateway-based setup today means the 3.8 upgrade is a config change later.

Bottom line

Qwen 3.8 is real, impressive by early accounts, and not yet a product you can build on. Evaluate it through the preview if the Token Plan pricing makes that cheap for you. Run production on Qwen 3.7-Max or open-weight Qwen in the meantime, and set up your stack so that swapping to 3.8 is a one-line change when the API or the weights actually arrive.

Run the current Qwen line through the Yotta AI Gateway, or check current GPU pricing if you'd rather hold the weights yourself. Full preview coverage in our Qwen 3.8-Max release breakdown.