model comparison

vs

Qwen4 Max vs Mistral Large 4

Qwen against Mistral AI, 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 Mistral Large 4 61.7%.
  • Mistral Large 4 lists lower input and output prices.
  • Qwen4 Max reads cached input for less ($0.075/M against Mistral Large 4's $0.136/M), which decides the bill for agents that resend long context.

side-by-side specs

specQwen4 MaxMistral Large 4
providerQwenMistral AI
input price$1.5/M$1.36/M
output price$4.5/M$4.18/M
cached input$0.075/M$0.136/M
blended price (3:1)$2.25/M$2.065/M
context window1M tokens1M tokens
max output262k tokens—
inputtext, images, videotext, images
structured outputyesyes
weightsclosedclosed
releasedNovember 2026October 6, 2026
DeepSWE v1.1 (maker-reported)75.761.7%

same work, both models

workloadQwen4 MaxMistral Large 4
one chat reply3,000 tokens in, 500 outQwen4 Max costs 1.1× as much as Mistral Large 4$0.0068$0.0062
summarise a 100-page report50,000 tokens in, 1,500 out, one request$0.08$0.07
1,000 classification calls500 tokens in, 20 out, per call$0.84$0.76
coding-agent session2M tokens in across ~60 turns, 60,000 out, no cache$3.27$2.97
same session, cached85% of input read from the prompt cache, the rest written to itMistral Large 4 costs 1.1× as much as Qwen4 Max$0.85$0.89

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.

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

Mistral Large 4

Mistral's mixture-of-experts model with about 1T total parameters (52B active): a hybrid instruct-and-reasoning model with a vision encoder and a 1M-token context. it entered public preview on 2026-10-06; Mistral says the open weights ship later.

pick it if

  • want a Mistral model that takes text and image input
  • need a 1M-token context with function calling and structured outputs
  • resend long prompts: cache hits bill $0.136 per 1M versus $1.36

skip it if

  • you need downloadable weights today; Mistral says they ship later
  • you need a stable release rather than a public preview

faq

which is better, Qwen4 Max or Mistral Large 4?

on DeepSWE v1.1, as each maker reports it, Qwen4 Max scores 75.7 and Mistral Large 4 61.7%. 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 Mistral Large 4?

Mistral Large 4 lists lower input and output prices. on a coding-agent session (2M tokens in, 60,000 out) Qwen4 Max costs $3.27 and Mistral Large 4 $2.97.

can i switch between Qwen4 Max and Mistral Large 4 mid-task?

yes. in an octomind session, /model octohub:qwen4-max and /model octohub:mistral-large-4 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