14×∞ days unlimited token giveaway: glm5.2/qwen35-122b-a10b/gpt-oss-120bgemma-3 and more — 14 days free trial with $400 credit, email verification only!

Business and Finance 羊毛党
1

Screenshot_20260712_143206
Screenshot_20260712_1432061080×1748 146 KB

As shown in the picture

It’s a service called Databricks, just select the domain email as shown

https://www.databricks.com/

Screenshot_20260712_142822
Screenshot_20260712_1428221080×1552 112 KB

API access is available: https://7474650858789139.ai-gateway.cloud.databricks.com/mlflow/v1
Here are some usable models (of course, any other open-source models would work, but Claude and ChatGPT seem to be unavailable):
system.ai.glm–5.2
system.ai.qwen35-122b-a10b
system.ai.gemma-3-12b
system.ai.gpt-oss-120b
(Of course there are others too, but I’m too lazy to test them, and I don’t know those models)

Here are the settings for GLM-5.2 (Databricks): Model glm-5-2, 128K context, output limit of 65536
Output speed is about 200 tokens/s, so in theory you could use up to 300 million tokens in a day (ideally, considering concurrency and minute restrictions)

Besides this, there are two similar services:

And Palantir (the big data and military-grade intelligence analysis giant) has its AIP (Artificial Intelligence developer platform, also provides Claude and GPTPlatform):

13 Likes
依旧是两个免费一个月，各种模型随便蹬的企业API
2

Where’s the address? What’s the point of giving a URL?

1 Like
3

The rates for popular models are now limited. With a trial account, it’s basically impossible to use popular models.

image
image1267×784 40.6 KB

4

image
image1363×829 105 KB

5

less than

6

How about trying out the competitor next door? Snowflake (seems to be its arch-rival): https://www.snowflake.com/en/product/features/cortex/

7

And there’s also Palantir (a giant in big data and military-grade intelligence analysis) with its AIP (Artificial Intelligence developer layer, which also offers Claude and gptPlatform) Palantir Artificial Intelligence Platform

1 Like
8

qwen3.5 122b works and it’s good enough :xhj10:

9

Let’s see what everyone else has to say first.

10

Is this the only one that’s available?

11

Then it might as well just use opencode’s free DeepSeekv4.

12

It seems like everything else is restricted, and the ones that are still usable are pretty lame :ac01:

13

Go into more detail.

14

With proxy pool rotation, you can achieve almost unlimited possibilities :xhj001:

15

Well, it’s still not as good as Nvidia.
But I did mention there are two other vendors — want to give them a try?

16

Can we milk it again?

17

system.ai.glm-5-2可以的，刚才测了
貌似就是所有开源模型都可以

18

来试试看

19

mmexport1783877176688
mmexport17838771766881485×867 8.46 KB

顺利蹬上glm5.2 :ac20:

20

Really easy to use, with basically no restrictions. The speed is nearly 200 tokens per second. As expected from the enterprise edition.