Gemini Flash 2.0 + Thinking (New and free)

Discussion
1

How to add Gemini Flash 2.0 and Gemini Flash 2.0 Thinking

Go to Google AI studio: https://aistudio.google.com
Get your free API key, add the key in Cursor, add the model in your list:

*gemini-2.0-flash-exp
*gemini-2.0-flash-thinking-exp

Free tier limit (might change)

  • Requests per minute (RPM): 15
  • Tokens per minute (TPM): 1 million
  • Requests per day (RPD): 1,500
  • Concurrent Sessions: 3 per API key

Bonus: You can also use Gemini Exp 1206 inside Google AI Studio with it’s full 2 million context window. It was able to solve issue I wasn’t able with other models.

Added new Thinking model. So far it seems excellent. o1 style and flash speed.

31 Likes
2

Hey,

Thanks for the feedback! We are constantly looking at the best models for us to add in to Cursor, and I appreciate the suggestion.

We have started to experiment with Gemini models, with one available in the app currently, so you may see more Gemini models available in the future.

1 Like
3

And please please don’t limit it to 30 k context when it can do 2 million

14 Likes
4

Indeed. @danperks please add it! You guys need to be fast. Adding 2 million context with Gemini Flash 2.0 would be SICK. Let us make mistakes! Let us be dumb! Give us freedom!

1 Like
5

add this with long context please :weary:

3 Likes
6

I think the longer context of the model is especially useful for a LARGE project understanding. But need to wait for Cursor team to test it completely.

2 Likes
7

At least read the API release announcement from Google , you need to wait until early 2025 to use the api , even with company like Cursor will need to have at least 1 month in order to adding it with some tweak and apply to cursor so you can wait for 2-3 month to use Gemini 2.0 with Cursor

8

At least read the API release announcement from Google , you need to wait until early 2025 to use the api

Is this really the case? I’m able to use gemini-2.0-flash-exp via Google Cloud Vertex AI via API as well.

9

At least read the API release announcement from Google , you need to wait until early 2025 to use the api , even with company like Cursor will need to have at least 1 month in order to adding it with some tweak and apply to cursor so you can wait for 2-3 month to use Gemini 2.0 with Cursor

It’s for the production ready Gemini 2.0, the Experimental one are available right now.

1 Like
10

can we also have it in composer agent? :pray::pray::pray:

1 Like
11

They should allow us to use our own model in composer agent when we are a pro user. Maybe the composer need way more testing and it could lead to unexpected result.

1 Like
12

I’d been using claude and gpt 4 all day to try to solve a problem and was not getting the right answer. When I switched to gemini it was solved in 2 minutes… halaleuah.

13

What version of Gemini

14

How do you create and use the 2.0-flash api key?
When I create a key and use it in cursor I get:
“The model gemini-2.0-flash-exp does not work with your current plan or api key”

1 Like
15

and please somebody explain why gemini models in cursor act like they’re not vision-capable?

and yeah, STILL not taking advantage of gemini’s already existing agentic capabilities, or 2M tokens is a bit of a downer…

16

Hi. The problem is that gemini-2.0-flash-exp in Cursor is only available in Chat, but not in Composer. Please fix this, I really want to test the new model on a large project

2 Likes
17

2.0, but now with the latest Cursor update it is telling me I can’t use it. Argh…

19

Be patient buddy

20

You can use the gemini-2.0-flash-exp model in the Roo Cline plug-in.
But after using it for two days, I feel that there is still a certain gap between it and Claude 3.5.
The advantage is that it is currently free to use

1 Like
21

I have added the new Gemini Flash Thinking, it seems very very good, let me know how it goes. But so far it follow the instruction better than sonnet 3.5. I don’t know how good at coding yet.