model comparison

vs

Qwen4 Max vs Qwen3.8 Max

two Qwen models, compared on listed price, real workload cost, context and capabilities.

the short answer

  • on DeepSWE v1.1, as each maker reports it, Qwen4 Max scores 75.7 and Qwen3.8 Max 56.6.
  • Qwen4 Max lists lower input and output prices.
  • Qwen4 Max reads cached input for less ($0.075/M against Qwen3.8 Max's $0.25/M), which decides the bill for agents that resend long context.

side-by-side specs

specQwen4 MaxQwen3.8 Max
providerQwenQwen
input price$1.5/M$2/M
output price$4.5/M$6/M
cached input$0.075/M$0.25/M
blended price (3:1)$2.25/M$3/M
context window1M tokens1M tokens
max output262k tokens131k tokens
inputtext, images, videotext, images, video
structured outputyesyes
weightsclosedopen (qwen3.8-max)
releasedNovember 2026August 2026
DeepSWE v1.1 (maker-reported)75.756.6

same work, both models

workloadQwen4 MaxQwen3.8 Max
one chat reply3,000 tokens in, 500 outQwen3.8 Max costs 1.3× as much as Qwen4 Max$0.0068$0.009
summarise a 100-page report50,000 tokens in, 1,500 out, one request$0.08$0.11
1,000 classification calls500 tokens in, 20 out, per call$0.84$1.12
coding-agent session2M tokens in across ~60 turns, 60,000 out, no cache$3.27$4.36
same session, cached85% of input read from the prompt cache, the rest written to itQwen3.8 Max costs 1.6× as much as Qwen4 Max$0.85$1.39

at listed rates. token counts are typical sizes, not measurements.

Qwen4 Max pricing notes

  • cache hits on the implicit context cache bill at $0.075 per 1M; cache writes bill at the normal input rate.

Qwen3.8 Max pricing notes

  • context caching is implicit: cache hits bill at $0.25 per 1M, cache writes at the normal input rate.

when to pick which

Qwen4 Max

Qwen4 Max is Alibaba's flagship Qwen4 model: a hosted 2.4T-parameter MoE that takes text, image and video, with a thinking mode. it succeeds Qwen3.8 Max at lower rates, and its weights are not published.

pick it if

  • you want Qwen4's flagship with image and video input
  • long agent sessions reuse a prefix; cache hits bill $0.075 per 1M, against $0.25 on Qwen3.8 Max
  • token-heavy agent runs; Alibaba reports 48 to 49% fewer tokens than Qwen3.8 Max on Terminal-Bench 4.0

skip it if

  • you need downloadable weights; Alibaba has announced open weights for smaller Qwen4 models, none for Max
  • you need audio input; Qwen3.8 Omni Flash takes audio, Max does not

Qwen3.8 Max

Qwen3.8 Max is Alibaba's flagship Qwen3.8 model: a hosted 2.4T-parameter MoE that takes text, image and video, with a thinking mode. it succeeds Qwen3.7 Max at lower rates and adds image input; its open weights are text-only.

pick it if

  • you want Qwen3.8's flagship with image and video input
  • long agent sessions reuse a prefix; cache hits bill $0.25 per 1M
  • you want a Max-class model whose text-only base weights are downloadable

skip it if

  • Qwen4 Max fits the task; same window and inputs, and it bills less on input, output and cache hits
  • Qwen3.8 Omni Flash covers the task; it takes images, video and audio, far cheaper

faq

which is better, Qwen4 Max or Qwen3.8 Max?

on DeepSWE v1.1, as each maker reports it, Qwen4 Max scores 75.7 and Qwen3.8 Max 56.6. these are each maker's own numbers, not a controlled head-to-head; the spec table lists every benchmark both publish.

which is cheaper, Qwen4 Max or Qwen3.8 Max?

Qwen4 Max lists lower input and output prices. on a coding-agent session (2M tokens in, 60,000 out) Qwen4 Max costs $3.27 and Qwen3.8 Max $4.36.

can i switch between Qwen4 Max and Qwen3.8 Max mid-task?

yes. in an octomind session, /model octohub:qwen4-max and /model octohub:qwen3.8-max switch models without losing the conversation, so you can draft on one and review on the other.

try either model

switch between models inside the same octomind session.

prices from the catalog of 2026-11-18