How much does the Qwen3.8 Max API cost?

There is no per-token API price — pay-as-you-go access has not been announced as of July 21, 2026. The only way to pay for qwen3.8-max-preview is a Token Plan subscription: Lite $6/month (700 credits per 5-hour window), Standard $18/month (3,000 credits), Pro $68/month (12,000 credits) at limited-time prices, with preview invocations consuming credits at as low as 10% of standard rates.

Is Qwen3.8 Max free to try?

There is no standalone free tier. The cheapest entry as of July 2026 is the $6/month limited-time Lite Token Plan, stretched further by the preview-period ~10% credit consumption rate.

Qwen3.8 Max Pricing: What's Real, What's Not Announced

A living tracker. Every number below is verified against Alibaba's official docs and carries its verification date — because several widely-shared numbers are already wrong.

There is no pay-as-you-go API price for Qwen3.8 Max. As of 2026-07-21, the only billed access is Alibaba's Token Plan subscription. During preview, qwen3.8-max-preview consumes credits at as low as 10% of standard rates.

Token Plan pricing (official, verified 2026-07-21)

Personal edition

Tier	List price	Limited-time price	Credits
Lite	$8/month	$6/month	700 / 5-hour window
Standard	$25/month	$18/month	3,000 / 5-hour window
Pro	$80/month	$68/month	12,000 / 5-hour window

Team edition (per seat)

Tier	List price	Limited-time price	Credits / month
Standard	$30	$20	25,000
Pro	$100	$75	100,000
Max	$200	$200	250,000

Correction watch: multiple news outlets report personal tiers as "$6 / $20 / $70". The official Token Plan page (checked 2026-07-21) lists $6 / $18 / $68 limited-time, against $8 / $25 / $80 list prices. If you're comparing costs, use the official numbers.

What a credit actually buys

Alibaba prices model invocations in credits, and the per-invocation credit cost varies by model. The load-bearing fact for Qwen3.8 Max: during preview, its consumption runs at as low as 10% of standard credit rates — Alibaba markets this as "10x capacity". Two practical implications:

Preview-period cost-per-task numbers you see anywhere (including ours, once we publish measurements) will not hold after the discount ends. Date-stamp everything.
Credits are quota-windowed (5-hour windows on personal tiers), so sustained heavy agent use hits window limits before monthly limits.

Not announced (don't trust numbers you see elsewhere)

Item	Status as of 2026-07-21
Pay-as-you-go per-token price	Not announced. Any $/M-token figure for Qwen3.8 Max is made up — for reference, rival Kimi K3 publishes $3/$15 per M tokens
GA / non-preview pricing	Not announced
Credit-to-dollar token equivalence	Not published; Token Plan credits cannot be honestly converted to a per-token price
Open-weights self-hosting cost	Weights promised, no date, no license — self-hosting economics unknowable today

Cheapest realistic ways to evaluate it

$6 Lite month + preview 10% consumption — enough for meaningful interactive coding trials.
Qoder / QoderWork — if you only want to feel out coding behavior.
For production budgeting today, price your workload on generally-available Qwen models instead — see the access guide for why the preview can't back a production app.

Running Qwen models in production? EvoLink's pay-as-you-go pricing covers 50+ models including the Qwen family, with no subscription. When Qwen3.8 Max gets a real API price, this page will compare routes the same day.

All prices verified against the official Token Plan page on 2026-07-21. Limited-time prices can end without notice; list prices shown alongside.

Sources: Token Plan overview (official) · Kimi K3 pricing context