qwen3.8-max API — per-token pricing and code examples

Checking statusPrice version 4 · in effect since Nov 6, 2026

Price per 1M tokens

InputOfficial price $2.00$0.20
Cache readOfficial price $0.25$0.02
Cache writeOfficial price $2.50$0.20
OutputOfficial price $6.00$0.60

Minimum per request$0.00002

Cache rates apply when the provider reports cache tokens.

90% off

About this model

qwen3.8-max is served through OpenAI Chat Completions, OpenAI Responses, Anthropic Messages. You pay $0.20 per 1M input tokens and $0.60 per 1M output tokens. That is 90% below the official price of $2.00 and $6.00. Cache reads cost $0.02 per 1M tokens when the provider reports cached tokens. It is included in budget packs from $9.99 up.

What Qwen3.8-Max is for

Alibaba Cloud describes Qwen3.8-Max as a 2.4-trillion-parameter MoE flagship and, in its press release, as the most powerful model in its Qwen series to date. Alibaba Cloud aims it at autonomous coding, office productivity, professional work in areas such as law, finance and design, and parsing long documents and videos. It was announced on August 3, 2026. Since September 5, 2026, Alibaba Cloud routes the qwen3.8-max name to the snapshot qwen3.8-max-0902, which it says keeps the 1M context window, thinking mode and the full tool ecosystem.

The model page lists a 1,000,000-token context. Input can be up to 991,808 tokens, or 983,616 in thinking mode, a reply can carry up to 131,072 output tokens, and the chain-of-thought itself can reach 262,144 tokens. Qwen3.8-Max accepts text, images and video as input and answers in text.

Hybrid thinking and how to bound it

Qwen3.8-Max uses hybrid thinking, and Alibaba Cloud turns thinking on by default. On the OpenAI-compatible Chat Completions API, enable_thinking switches it, and reasoning_effort takes low, medium or xhigh, with xhigh as the default. Alibaba Cloud warns that reasoning_effort and thinking_budget cannot be set in the same request; sending both causes an error.

Output limits also need care, because the meaning of max_tokens changes with the API format. On Alibaba Cloud's Chat Completions API, max_tokens caps only the answer and excludes the chain-of-thought, while max_completion_tokens caps thinking plus answer together. On Alibaba Cloud's own Anthropic-compatible API, max_tokens for Qwen3.8-Max covers both, according to Alibaba Cloud's documentation.

Multi-turn conversations have their own rule. preserve_thinking defaults to true for this model, and Alibaba Cloud asks you to return earlier reasoning in the reasoning_content field, never pasted into content. For long prompts with thinking on, Alibaba Cloud recommends a client timeout of 180 seconds or more.

{
  "model": "qwen3.8-max",
  "reasoning_effort": "medium",
  "max_completion_tokens": 32000,
  "messages": [{ "role": "user", "content": "Review this contract clause and list the risks" }]
}

That is the /v1/chat/completions format. On Alibaba Cloud's own Anthropic-compatible API the depth setting is output_config.effort, where max and high are mapped to xhigh.

What it costs on Cent AI

Qwen3.8-Max is billed per token from your prepaid USD balance at the four rates in this page's price box. Thinking is billed as output, and with the default xhigh effort and room for 262,144 thinking tokens, a single hard request can produce far more output than its visible answer. Because preserved reasoning from earlier turns is sent back, Alibaba Cloud counts that history as input tokens on the next turn, so long conversations grow in cost from both sides.

Alibaba Cloud offers two cache modes. Implicit caching is automatic and cannot be disabled, but hits are not guaranteed. Explicit caching uses cache_control with {"type": "ephemeral"}, needs at least 1,024 tokens, and stays valid for 5 minutes, reset on each hit. Cent AI bills cache reads and writes only from the cached tokens Alibaba Cloud reports in usage.

Using it from your tools

Qwen3.8-Max can be called on /v1/chat/completions, /v1/responses and /v1/messages. Cent AI does not change your parameters silently. Tool calling and structured outputs are supported; on Chat Completions, response_format with json_object needs the word JSON in your prompt, while json_schema does not, and Alibaba Cloud recommends strict: true. Parallel tool calls are on by default.

Specification

Context window
—
Input and output
Text
Endpoints
/v1/chat/completions, /v1/responses, /v1/messages
Smallest pack
$9.99

Latency and error rate per model are on the status page. status.cent-ai.dev

What a request costs

Open the full calculator

Cost per request

90% is an assumption you can change, not a promise. Cache is billed only from cache tokens the model provider reports.

See how many such requests it covers.

Estimate

qwen3.8-max
Fresh input99,000 tok
Cache read891,000 tok
Output10,000 tok

Per request$0.0436
At official pricesAt official prices $0.4807

Total$0.0436

An estimate at price version 4. Your bill comes from the usage the provider reports for each request.

Save $0.4371

Code

curl https://api.cent-ai.dev/v1/chat/completions \
  -H "Authorization: Bearer $CENT_AI_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "qwen3.8-max", "messages": [{"role": "user", "content": "Hello"}]}'

Set CENT_AI_KEY to a key from the dashboard. OpenAI SDKs use https://api.cent-ai.dev/v1; Anthropic SDKs use https://api.cent-ai.dev.

Price history

Version 4, in effect since Nov 6, 2026.

  1. Effective

    per 1M tokens

    Inputold $0.10 new $0.20
    Cache readold $0.01 new $0.02
    Cache writeold $0.10 new $0.20
    Outputold $0.30 new $0.60
  2. Effective

    per 1M tokens

    Inputold $0.10 new $0.10
    Cache readold $0.01 new $0.01
    Cache writeold $0.125 new $0.10
    Outputold $0.30 new $0.30
  3. Effective

    per 1M tokens

    Inputold $0.10 new $0.10
    Cache readold $0.0125 new $0.01
    Cache writeold $0.125 new $0.125
    Outputold $0.30 new $0.30

Compare with

All models against official prices

Questions

How much does qwen3.8-max cost on Cent AI?

$0.20 per 1M input tokens and $0.60 per 1M output tokens. Cache reads cost $0.02 and cache writes $0.20 per 1M tokens. Each request is rounded up once.

Is there a minimum charge?

A successful request costs at least $0.00002. A request that ends early, because the client left or the output was cut off, pays only for its tokens.

How do I pay?

Top up a prepaid USD balance with QRIS or USDT/USDC, or buy a budget pack once. Nothing renews automatically.

What if the provider does not report usage?

The request is billed from an estimate and marked estimated on the Requests page. An estimate never contains cache tokens.

Which endpoint do I call?

/v1/chat/completions, /v1/responses, /v1/messages. The OpenAI base URL is https://api.cent-ai.dev/v1.

Try qwen3.8-max with one key

One prepaid balance for every model. Pay with QRIS or USDT/USDC.

Start free