Zhipu AI language model

GLM-5.3 Flash

the first natively multimodal model in Zhipu AI's GLM-5 series, an open-weights 320B-total, 18B-active sibling of GLM-5.3. it costs about a ninth of GLM-5.3's per-token rate and, unlike GLM-5.3, accepts image and video input.

01

price

$0.15/M

input tokens

$0.5/M

output tokens

$0.03/M

cached input

per 1M tokens. octomind bills the listed provider price, with no markup, on every plan, including the free one (pay as you go from credits).

  • Z.ai's 50% promo ($0.075 / $0.25) ended 2026-09-09 16:00 UTC; the listed price is the standard rate.

02

what real work costs

workloadcost
one chat reply 3,000 tokens in, 500 out $0.0007
summarise a 100-page report 50,000 tokens in, 1,500 out, one request $0.0083
1,000 classification calls 500 tokens in, 20 out, per call $0.09
coding-agent session 2M tokens in across ~60 turns, 60,000 out, no cache $0.33
same session, cached 85% of input read from the prompt cache, the rest written to it $0.13

at the listed rate. token counts are typical sizes, not measurements.

03

specs

provider
Zhipu AI
context window
1M tokens — ≈ 1,500 pages of prose or 100,000 lines of code
max output
128k tokens
input
text, images, video
output
text
structured output
yes — JSON schema
reasoning mode
yes
prompt caching
yes — cache hits at $0.03/M
weights
open — MIT
released
August 2026
status
live
model id
glm-5.3-flash

integration notes

  • thinking.type only accepts enabled; requests that send disabled are not supported.
  • reasoning_effort supports only low, high or max on GLM-5.3-Flash; it defaults to max.

04

benchmarks

as published by the maker; not measured by octomind.

benchmarkscore
Terminal Bench 2.1source84.3
DeepSWE v1.1source63.4
HLE with Toolssource55.3
Toolathlon Verifiedsource78.4
MMVUsource80.5

05

when to pick it

pick it if

  • you want GLM-5.3-family text at roughly a ninth of the price
  • you need image or video input in the GLM line
  • you want MIT-licensed weights you can self-host

06

where it sits on price

blended price $0.237/M at 3 input tokens per output token: 9 live language models cost less, 34 cost more.

  1. GPT-6 Luna $0.2/M
  2. Qwen3.8 Omni Flash $0.23/M
  3. GLM-5.3 Flash $0.237/M
  4. MiniMax M3.1 Flash $0.262/M
  5. GPT-OSS 120B $0.45/M

07

use GLM-5.3 Flash in octomind

one octohub login reaches every model in the catalog. pick this one per run, per session, or as your default.

run it
octomind login
octomind run -m octohub:glm-5.3-flash
switch mid-session
/model octohub:glm-5.3-flash
make it the default — config.toml
[model]
name = "octohub:glm-5.3-flash"

08

compare GLM-5.3 Flash

all model comparisons →

09

similar models

nearest live models by listed input price.

view all models →

10

GLM-5.3 Flash faq

how much does GLM-5.3 Flash cost?

GLM-5.3 Flash lists at $0.15/M input, $0.5/M output, $0.03/M cached input per 1M tokens. a typical coding-agent session (2M tokens in, 60,000 out) comes to $0.33, or $0.13 when 85% of the input is read from the prompt cache. octomind bills the listed provider price, with no markup, on every plan, including the free one (pay as you go from credits).

what is the context window of GLM-5.3 Flash?

1M tokens, ≈ 1,500 pages of prose or 100,000 lines of code. a single reply can run to 128k tokens.

does GLM-5.3 Flash accept images?

yes. it takes text, images and video as input and returns text.

is GLM-5.3 Flash open source?

its weights are open under MIT, so it can also be self-hosted.

when was GLM-5.3 Flash released?

August 2026, by Zhipu AI.

is GLM-5.3 Flash free?

no. GLM-5.3 Flash bills per token at $0.15/M input, $0.5/M output, $0.03/M cached input per 1M. octomind's free plan reaches it pay as you go from credits, so you pay only for the tokens you use, at the provider price. credits carry a 3% top-up fee.

how do i use GLM-5.3 Flash in octomind?

run octomind login once, then octomind run -m octohub:glm-5.3-flash. inside a running session, /model octohub:glm-5.3-flash switches to it without losing context.

use GLM-5.3 Flash in octomind

choose a model when you start, then switch mid-session when the work changes.