uttapen

Qwen3.8 2.4T A95B (batch) API: toman pricing and code

qwen/qwen3.8-2.4t-a95b:batch

toolsreasoningjson
Input · per 1M tokens
580,250 toman
Output · per 1M tokens
1,740,750 toman
One 1,000-word request ≈
3,017 toman

You are billed for the usage the request actually reported. Prices follow the market. Cached input: 72,531 toman / 1M.Pricing and top-ups

Qwen3.8 2.4T A95B (batch) example: JSON output against a schema

The example is picked from this model's own capabilities. Drop in your key and it runs as is.

from openai import OpenAI
import json

client = OpenAI(base_url="https://api.uttapen.ir/v1", api_key="sk-up-…")

schema = {
    "name": "ticket",
    "schema": {
        "type": "object",
        "properties": {
            "category": {"type": "string", "enum": ["fani", "mali", "forush"]},
            "priority": {"type": "integer", "minimum": 1, "maximum": 5},
            "summary": {"type": "string"},
        },
        "required": ["category", "priority", "summary"],
        "additionalProperties": False,
    },
    "strict": True,
}

resp = client.chat.completions.create(
    model="qwen/qwen3.8-2.4t-a95b:batch",
    messages=[{"role": "user", "content": "Ticket: "For two days I cannot download my invoice and I was charged twice.""}],
    response_format={"type": "json_schema", "json_schema": schema},
)
print(json.loads(resp.choices[0].message.content))

What is Qwen3.8 2.4T A95B (batch) good for?

The id for Qwen3.8 2.4T A95B (batch) in our API is "qwen/qwen3.8-2.4t-a95b:batch", served from Qwen (Alibaba). Context is 1,010,000 tokens; past that you have to summarise the history yourself. A single response can run to 909,000 tokens. On price it sits in the "mid-range" band — cheaper than 297 and dearer than 109 of the other paid models in the catalogue.

What it can do beyond plain text: it supports tool calling, so it can invoke your own functions with valid arguments; it returns schema-valid JSON through response_format, ready to hand to your code; it has a reasoning mode that pays off on multi-step problems, maths and debugging. All of it works through the standard parameters of the official OpenAI SDK — no custom client, no wrapper. Keep in mind that reasoning tokens are output tokens and do appear on the bill.

Pricing is 580,250 toman per 1M input tokens and 1,740,750 per 1M output tokens. A 1,000-word round trip on Qwen3.8 2.4T A95B (batch) lands near 3,017 toman. It supports cached input: repeated context is billed at 72,531 toman per 1M, which matters a lot if your system prompt is long. You are always charged for the usage the request actually reported, never for the estimate, and a failed request costs nothing.

The closest alternative with the same capabilities from a different provider is Aion-3.0: Qwen3.8 2.4T A95B (batch) works out roughly 1.1× cheaper, and its context window is larger. Both run on the same key and the same code, so trying the other one is a single string change.

To make the figure concrete: 100,000 toman of credit buys roughly 33 thousand-word requests on Qwen3.8 2.4T A95B (batch), and every 1,000 toman is about 331 words of round trip. A job with one million input tokens and one million output tokens comes to 2,321,000 toman in total. Filling this model's 1,010,000-token window costs 586,053 toman on the input side alone, which is the real reason to keep conversation history short.

This id gets confused with "qwen/qwen3.8-2.4t-a95b", because the underlying model is the same. The difference is the suffix: "batch" against the standard variant. Both land at nearly the same price on a thousand-word request (3,017 and 3,017 toman). The context windows differ too: 1,010,000 against 1,048,576 tokens. Maximum answer length differs as well: 909,000 against 262,144 tokens. On parameters the other takes seed.

Qwen (Alibaba) has 53 models in our catalogue; the cheapest is Qwen3.7 Flash at 60 toman per thousand words and the dearest Qwen3.8 Max (0902) at 3,017. Among the less common parameters it accepts logit_bias, logprobs, min_p, reasoning_effort, repetition_penalty, top_k — all through the standard request body. It does not support seed, which most models here do, so test before switching if your code relies on them. By context size the nearest option from another provider is MiniMax-01 at 1,000,192 tokens.

Good fits: product chatbots and internal assistants, where cost and quality have to balance, logic puzzles and code review, agents that reach out to APIs and databases, extracting data against a fixed schema, analysing a long document or codebase in one request.

Provider's own description

Qwen3.8 2.4T A95B is an open-weight sparse mixture-of-experts model from Qwen and the open-weight variant of [Qwen3.8 Max](/qwen/qwen3.8-max), with 95 billion active parameters out of 2.4 trillion total. It is...

Three real jobs, priced on this model

Each figure is derived from the prices above and moves when they do.

JobTokensCost
One chat turn with a medium history1,500 in + 400 out1,567 toman
Summarising a ten-page document4,000 in + 600 out3,365 toman
Classifying a thousand short rows120,000 in + 20,000 out104,445 toman

Other variants of this model

Same core model, different execution terms and different price. This page is the batch (cheaper, slower) variant.

VariantModel idOutput / 1MContext
standardqwen/qwen3.8-2.4t-a95b1,740,7501,048,576

Put the variant's id verbatim in the model field; nothing else in your code changes.

Frequently asked

How do I call Qwen3.8 2.4T A95B (batch) from Iran?
Sign up with your mobile number, top the wallet up in toman, create an API key, then in the official OpenAI SDK point base_url at https://api.uttapen.ir/v1 and set model to "qwen/qwen3.8-2.4t-a95b:batch". Nothing else in your code changes.
What does Qwen3.8 2.4T A95B (batch) cost in toman?
580,250 toman per 1M input tokens and 1,740,750 toman per 1M output tokens; a 1,000-word request is around 3,017 toman. You pay for the usage that request actually reported, and a failed request costs nothing.
How much input does Qwen3.8 2.4T A95B (batch) take?
Up to 1,010,000 tokens per request, roughly 758k words. A single answer can reach 909,000 tokens.
Does Qwen3.8 2.4T A95B (batch) support streaming and tool calling?
Streaming (stream=true) works on every model here. This one supports tool calling in the standard OpenAI shape. Structured output through response_format works too.