The GLM-5 family
GLM-5.2 and GLM-5.1 cost exactly the same, $1.4 in and $4.4 out, but the 5.2 context window is one million tokens against 200K on 5.1. So the version to use is 5.2, and for a new build 5.1 has no reason to exist.
- four members on the price list
- 5.2 window is five times 5.1
- the price rose along the line
Last checked: This is a reference page. It is re-checked against the vendor sources and updated when a new version ships. What changed
Which version to use
GLM-5.2. The Z.ai price table gives 5.1 and 5.2 one figure rather than two close ones: $1.4 in and $4.4 out. Both guide pages also put max output at 128K tokens. The difference sits in a single column, the context window, and there 5.1 stops at 200K while 5.2 reaches one million. When two models charge the same and one accepts five times the input, the choice stops being a decision.
A price that rose along the line
GLM-5 opened at $1 in and $3.2 out. Then 5.1 and 5.2 arrived at $1.4 and $4.4, which is forty percent more on input and about thirty-seven percent more on output. That is the inverse of the pattern we wrote up on the Anthropic Opus family, where the price did not move across five consecutive versions.
For anyone running a service the practical translation is this: moving from GLM-5 to 5.2 raises the invoice, and it should not be mistaken for a free upgrade. What the same move buys is five times the window. Not a bad trade, but a trade.
Turbo is not what its name suggests
GLM-5-Turbo sits between 5 and 5.1 at $1.2 and $4, and the name implies one thing: it is faster. Its own page says something else. It presents the model as a "ClawBench Enhanced Model" that is "deeply optimized for the OpenClaw scenario", and there is no latency or throughput figure anywhere on that page. Its context window is the same 200K as GLM-5, while it charges twenty percent more.
So buying Turbo for speed means buying something the vendor never sold as fast. If you work inside OpenClaw the story changes, and it becomes the only member of the family making a specific claim about your job.
A timeline with no dates in it
The order of these four versions is clear from the documentation index. The dates are not. None of the Z.ai guide pages carries a release date, so nobody can say how many months separate 5 from 5.2, or how long any version has been in service. For a section that dates everything, that gap is genuinely annoying.
The same documentation also leaks a staleness tell. The 5.1 page still calls itself the latest flagship model at Z.ai, while 5.2 carries the HOT label in the index and is described as a flagship on its own page. Both pages are live today and they contradict each other.
Release timeline
- GLM-5.2 Max current The second cheapest output price in the coding table, behind DeepSeek only.
Using it from Iran
This column carries a date and says how it was checked, because most listicles guess it. Where we have only read a vendor policy page, the note below says exactly that.
- Reachable
- not checked
- Payment
- not checked
- Free tier
- no
How we checked: Z.ai publishes no supported-countries list, so we can write neither open nor blocked. We have not tested from an Iranian connection either.
What it is good at
- The one million token window on GLM-5.2, at the price of the 200K version
- Cached input at $0.26 on 5.2, roughly a fifth of the standard rate
- Four versions are available at once, so no migration is forced
- The cheapest member of the family still sells at $1 input
Where it falls short
- Z.ai publishes no release date for any of these four versions, so this family has an order and not a timeline.
- The price rose along the line, so moving from GLM-5 to 5.2 increases the invoice.
- The price table marks cached input storage as free for a limited time, which means a figure that is zero today is scheduled to change.
- Z.ai publishes no comparative benchmark between 5.1 and 5.2, so the difference in answer quality is not measurable and only the window is.
- Access from Iran is unchecked, and Z.ai publishes no country list.
Our take
For a new build take GLM-5.2 and do not even look at 5.1, because they cost the same. If you already run a service on GLM-5, work the forty percent input gap out against your own real volume before migrating, not against the unit price.
Questions people actually ask
How much better is GLM-5.2 than 5.1
On paper the main difference is the context window: one million tokens against 200K. Z.ai publishes no comparative benchmark between the two, so any claim about answer quality between these versions is unsourced. Their prices are identical.
Is GLM-5-Turbo faster
Its own page never says so. It presents the model as a "ClawBench Enhanced Model" optimised for the OpenClaw scenario and gives no latency or speed figure at all. Its context window is the same 200K as GLM-5, while it costs twenty percent more.
Sources
- Z.ai pricing vendor source 12 August 2026
- Z.ai GLM-5.2 guide vendor source 12 August 2026
- Z.ai GLM-5.1 guide vendor source 12 August 2026
- Z.ai GLM-5 guide vendor source 12 August 2026
- Z.ai GLM-5-Turbo guide vendor source 12 August 2026