model comparison

vs

Mistral Large 4 vs GLM-5.3

Mistral AI against Zhipu AI, compared on listed price, real workload cost, context and capabilities.

the short answer

  • on DeepSWE v1.1, as each maker reports it, Mistral Large 4 scores 61.7% and GLM-5.3 66.9.
  • Mistral Large 4 lists lower input and output prices.
  • Mistral Large 4 supports vision; GLM-5.3 does not.
  • Mistral Large 4 reads cached input for less ($0.136/M against GLM-5.3's $0.26/M), which decides the bill for agents that resend long context.

side-by-side specs

specMistral Large 4GLM-5.3
providerMistral AIZhipu AI
input price$1.36/M$1.4/M
output price$4.18/M$4.4/M
cached input$0.136/M$0.26/M
blended price (3:1)$2.065/M$2.15/M
context window1M tokens1M tokens
max output—128k tokens
inputtext, imagestext
structured outputyesyes
weightsclosedopen (GLM-5.3 License (MIT-style; Model-as-a-Service operators above $10B revenue must pass a Z.AI security review))
releasedOctober 6, 2026August 14, 2026
DeepSWE v1.1 (maker-reported)61.7%66.9

same work, both models

workloadMistral Large 4GLM-5.3
one chat reply3,000 tokens in, 500 outwithin 5%$0.0062$0.0064
summarise a 100-page report50,000 tokens in, 1,500 out, one request$0.07$0.08
1,000 classification calls500 tokens in, 20 out, per call$0.76$0.79
coding-agent session2M tokens in across ~60 turns, 60,000 out, no cache$2.97$3.06
same session, cached85% of input read from the prompt cache, the rest written to itGLM-5.3 costs 1.3× as much as Mistral Large 4$0.89$1.13

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

GLM-5.3 pricing notes

  • cache hits bill at $0.26 per 1M input tokens; there is no separate cache-write charge.

when to pick which

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

GLM-5.3

GLM-5.3 is Z.ai's open-weight, text-only reasoning model for coding and long-horizon tasks. it keeps GLM-5.2's base model and price, with its gains coming from post-training, and reads up to 1M tokens of context.

pick it if

  • you want open weights for coding and long-horizon tasks
  • you work with long inputs: 1M context, cache hits at $0.26 per 1M
  • you need JSON structured output from a reasoning model

skip it if

  • you need image input; GLM-5.3 Flash reads images for far less
  • you want answers without reasoning; thinking cannot be switched off

faq

which is better, Mistral Large 4 or GLM-5.3?

on DeepSWE v1.1, as each maker reports it, Mistral Large 4 scores 61.7% and GLM-5.3 66.9. these are each maker's own numbers, not a controlled head-to-head; the spec table lists every benchmark both publish.

which is cheaper, Mistral Large 4 or GLM-5.3?

Mistral Large 4 lists lower input and output prices. on a coding-agent session (2M tokens in, 60,000 out) Mistral Large 4 costs $2.97 and GLM-5.3 $3.06.

which one accepts images?

Mistral Large 4 supports vision; GLM-5.3 does not.

can i switch between Mistral Large 4 and GLM-5.3 mid-task?

yes. in an octomind session, /model octohub:mistral-large-4 and /model octohub:glm-5.3 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