Last verified
200 TOKENS/S2.5X GLM-5.3-FLASHSAME MODEL AS FLASHIMAGE + VIDEO IN1M CONTEXT

GLM-5.3-FlashX API Pricing

GLM-5.3-FlashX is GLM-5.3-Flash served for speed. Z.ai's model page covers both ids with one spec - image, video, file and text in, 1M context, thinking always on - and says FlashX runs at 200 tokens per second. The price for that is $0.37/M input, $0.075/M cached input and $1.25/M output: exactly 2.5x Flash on cached input and output, and 2.47x on input. Z.ai published no release date for it, and it is not yet on the GLM Coding Plan.

Input - per 1M tokens
$0.37/M
GLM-5.3-Flash $0.15/M vs Flash
Output - per 1M tokens
$1.25/M
GLM-5.3-Flash $0.50/M thinking billed here
Cached input - per 1M tokens
$0.075/M
Cache storage free for now -80%
Effective - agentic blend
$0.218/M
92/8 split - 82% cache hit rate
§ 01 / TERMINAL

Run the numbers.

Calculator pre-loaded with GLM-5.3-FlashX's International rates. Every result is about 2.5 times what the same tokens cost on GLM-5.3-Flash; the difference buys output speed, not a different model. Thinking cannot be turned off, and its tokens bill as output.

$ /mo
Workload split
Prompt cache hit rate
Tokens you can process
—
Words equivalent (English)
—
Effective rate
—
Open full calculator (all models · share URL · CSV) →
§ 02 / SCENARIOS

Real-world presets.

CODING AGENT

Repo-wide bug fix

$0.028/task
120,000 in - 10,000 out~3,587 units/$100
UI FROM SCREENSHOT

Page rebuilt from a mockup

$0.015/page
40,000 in - 8,000 out~6,612 units/$100
DOCUMENTS

Report from a file bundle

$0.013/report
60,000 in - 4,000 out~7,882 units/$100
ASSISTANT

Interactive chat turn

$0.00177/turn
6,000 in - 800 out~56,541 units/$100
§ 03 / TOKENIZER

Paste text. See tokens. See cost.

What's a token →
Calibrated · measured on the vendor's tokenizer · 2026-06-10 Auto-counts as you type

Counts use a chars-per-token calibration measured on the vendor's own published tokenizer (zai-org/GLM-5, 2026-06-10). English prose is typically within a few percent; code and non-Latin scripts tokenize heavier. For billing-exact counts use the vendor's count-tokens API.

Characters 364
Words 59
Tokens (estimated) 69 tokens
Cost as input · uncached $0.00003 USD
Cost as output · uncached $0.00009 USD
Cost as cached input $0.00001 USD
§ 04 / SHELF

Up against the shelf.

All models →
Model Input /M Output /M Effective blended Context Best for
GLM-5.3-FlashX Current $0.37 cache $0.075 $1.25 $0.218 agentic 92/8 1M GLM-5.3-Flash where output speed matters
GLM-5.3-Flash $0.15 cache $0.03 $0.50 $0.0875 −2.5× 1M The same model at 40% of the price
GLM-5.3 $1.40 cache $0.26 $4.40 $0.78 +3.6× 1M The full-size flagship, text only
GLM-4.7 FlashX $0.07 cache $0.01 $0.40 $0.0511 −4.3× 200K Cheapest paid GLM, no multimodal
MiMo-V2.6-Flash $0.14 cache $0.0028 $0.28 $0.0477 −4.6× 1M Omni-modal input at a lower rate
DeepSeek V4.1 Flash $0.30 cache $0.006 $1.20 $0.15 −31% 1M Cheap 1M-context agent loops with images
Gemini 3.8 Flash $0.75 cache $0.075 $3.75 $0.481 +2.2× · promo 1M The Western flash tier

Frequently asked.

What GLM-5.3-FlashX costs, what the extra money buys over GLM-5.3-Flash, and what Z.ai has not published.

Q · 01 How much does GLM-5.3-FlashX cost? +
$0.37/M input, $0.075/M cached input and $1.25/M output on Z.ai's International price table, or $0.22/M on our 92/8 agentic blend with 82% cache hits. Cached-input storage is listed as limited-time free, a separate charge this page does not model.
Q · 02 Is it a different model from GLM-5.3-Flash? +
No. Z.ai documents the two on one page, GLM-5.3-Flash/FlashX, with one set of specs, and describes FlashX as delivering inference speeds of 200 tokens per second. It has its own model code, glm-5.3-flashx, and its own price, which is why it has its own page. Z.ai does not publish a speed figure for plain Flash, so the size of the speed-up is not stated.
Q · 03 How much more does it cost than GLM-5.3-Flash? +
Exactly 2.5x on cached input and output at the time of writing, and 2.47x on input: GLM-5.3-Flash lists at $0.15 / $0.03 / $0.50. On our blend that makes FlashX roughly two and a half times the cost per task. Pay it where a person is waiting on the output; for batch work, background agents and anything read later, plain Flash gives the same answers for 40% of the price.
Q · 04 When was it released? +
Z.ai has not said. Its release notes date GLM-5.3-Flash to August 26, 2026 and do not mention FlashX; Internet Archive captures of the model page through September 5 carry no FlashX, and OpenRouter listed it on September 18. The release date shown on this page is the day we first verified the price on Z.ai's own table, and it will be corrected if Z.ai publishes one.
Q · 05 What does it accept, and can thinking be turned off? +
Images, video, files and text in; text out; a 1M-token context and up to 128K output tokens. Thinking cannot be disabled - thinking.type only accepts enabled - so reasoning tokens always bill at the output rate. Function calling, structured output and context caching are supported.
Q · 06 Can I use it on the GLM Coding Plan? +
Not yet. Z.ai's model page says GLM-5.3-Flash is fully available on the GLM Coding Plan and that FlashX is not yet on it. Until that changes, FlashX is pay-as-you-go only at the rates on this page.
Y
Reviewed by Yaroslav Vikhariev Founder · AI//COST · Prices checked by hand against docs.z.ai · Last verified Sep 24, 2026
Methodology Report a correction More by Y.V.