Qwen3.8 2.4T A95B

Provider: Alibaba (Qwen) CN Official API price

Input price
$2.00
/ 1M tokens
Output price
$6.00
/ 1M tokens
Cache price
$0.25
/ 1M tokens (read)
Context
1M tokens
Max output 131K tokens

Overview

Alibaba's Qwen3.8 2.4T A95B is an open-weight sparse MoE flagship featuring 95B active parameters out of 2.4T total. It handles heavy reasoning, function calling, and JSON workloads across a 1M context window.

Strengths

  • 95B active parameters from 2.4T total MoE architecture
  • 1M token context window with 131k max output
  • Native support for thinking, function-calling, and JSON mode
  • Cached input discounted to $0.25 per million tokens

Best use cases

  • Complex reasoning and automated multi-step agent workflows
  • Parsing and generating structured JSON over massive contexts
  • Large-scale document analysis within the 1M window

Pricing position

Input is $2/M and output is $6/M, matching Mistral Large 2407 while beating GPT-4.1's $8 output rate.

Caveats

Carries no free tier and runs significantly pricier than Qwen's own Flash and 235B variants.

Notes drafted by an LLM from the figures on this page (2026-11-10) and rewritten when the price changes.

Key facts

Price basis
Official API price
Max output
131K tokens
Listed on OpenRouter
2026-08-12
Tracked since
2026-08-12
Last updated
2026-11-10

Features

Function callingJSON outputReasoning

Price history

  • 2026-08-12 Added

Monthly cost estimate

Scale Input (M) Output (M) Monthly
0.1M (hobby) 0.1 0.05 $0.50
1M (small app) 1 0.5 $5.00
10M (mid scale) 10 5 $50.00
100M (large) 100 50 $500.00

※ Cache and batch discounts not factored in. Detailed calculator

Official resources

Other models from Alibaba (Qwen)

Similar-priced models

Source: OpenRouter API (qwen/qwen3.8-2.4t-a95b)