I recently set up an n8n workflow for automation. The goal was to scrape my blog posts from Naver, automatically translate them into multiple languages, and publish them to my WordPress site. It worked perfectly a few times, but then it suddenly stopped, throwing errors. After tracing the cause, I discovered that the Gemini API was rejecting my requests.
In a panic, I checked the Google AI Studio dashboard. The RPD (Requests Per Day) section was glowing red. Under the Gemini API free plan, both Gemini 2.5 Flash and 3.5 Flash have strict limits: 5 requests per minute and a total of 20 per day. Since my workflow runs 1 main translation plus 3 loop translations per post, I was hitting the daily limit after processing just five or six posts.
Recommended Gemini API Free Plan Models
| Model | Category | RPM | TPM | RPD |
| :— | :— | :— | :— | :— |
| Gemini 2.5 Flash | Text | 5 | 250K | 20 |
| Gemini 3.5 Flash | Text | 5 | 250K | 20 |
| Gemini 3.1 Flash Lite | Text | 15 | 250K | 500 |
| Gemini 2.5 Pro | Text | 0 | 0 | 0 |
| Gemini 2 Flash | Text | 0 | 0 | 0 |
| Gemini 2 Flash Lite | Text | 0 | 0 | 0 |
| Gemini 2.5 Flash TTS | Multimodal | 3 | 10K | 10 |
| Gemini 2.5 Pro TTS | Multimodal | 0 | 0 | 0 |
| Imagen 4 Generate | Multimodal | – | – | 25 |
| Imagen 4 Ultra Generate | Multimodal | – | – | 25 |
| Imagen 4 Fast Generate | Multimodal | – | – | 25 |
| Gemma 4 26B | Other | 15 | Unlimited | 1.5K |
| Gemma 4 31B | Other | 15 | Unlimited | 1.5K |
| Gemini Embedding 1 | Other | 100 | 30K | 1K |
| Gemini 3.1 Pro | Text | 0 | 0 | 0 |
| Gemini 2.5 Flash Lite | Text | 10 | 250K | 20 |
| Nano Banana (Gemini 2.5 Flash Preview) | Multimodal | 0 | 0 | 0 |
| Gemini 3 Flash | Text | 5 | 250K | 20 |
| Nano Banana Pro (Gemini 3 Pro) | Multimodal | 0 | 0 | 0 |
| Nano Banana 2 (Gemini 3.1 Flash) | Multimodal | 0 | 0 | 0 |
| Lyria 3 Clip | Multimodal | 0 | 0 | 0 |
| Lyria 3 Pro | Multimodal | 0 | 0 | 0 |
| Veo 3 Generate | Multimodal | 0 | – | 0 |
| Veo 3 Fast Generate | Multimodal | 0 | – | 0 |
| Veo 3 Lite Generate | Multimodal | 0 | – | 0 |
| Gemini 3.1 Flash TTS | Multimodal | 3 | 10K | 10 |
| Gemini Robotics ER 1.5 Preview | Other | 10 | 250K | 20 |
| Gemini Robotics ER 1.6 Preview | Other | 5 | 250K | 20 |
| Computer Use Preview | Other | 0 | 0 | 0 |
| Gemini Embedding 2 | Other | 100 | 30K | 1K |
| Antigravity | Agent | 0 | 0 | 0 |
| Deep Research Pro Preview | Agent | 0 | 0 | 0 |
I had no choice but to scour the free plan for other available models. There aren’t many options, so the choice was easy. There are two models that offer great value: Gemma 4 31B and Gemini 3.1 Flash Lite. Both allow for 500 to 1,500 daily requests, which is perfect.
Gemma 4 31B
- Generous Limit: With 1,500 requests per day, it handles 3 times more work than Gemini 3.1 Flash Lite. Given that my workflow requires 4 calls per execution, I can run it up to 375 times a day.
- No Token Limit: Since the Tokens Per Minute (TPM) limit is unlimited, you don’t have to worry about the connection dropping due to long blog posts or many image URLs.
- Output Characteristics: It may include reasoning processes or unnecessary tags in its output, so you must strictly instruct the prompt to output only the translated result to minimize post-processing.
Gemini 3.1 Flash Lite
- Instruction Adherence: It tends to output clean Markdown and text as requested within the Google API environment, requiring fewer exceptions.
- Speed: As a low-latency, lightweight model, it offers fast processing speeds.
- Capacity Limit: With a daily limit of 500 requests (up to 125 workflow executions), you need to monitor your cumulative limits if you are publishing a large volume of content.
Which Gemini API Model for n8n?
If you are publishing fewer than 100 posts a day and want clean Markdown output without extra formatting, Gemini 3.1 Flash Lite is the better choice. Conversely, if you have high daily volume and want to avoid “token exceeded” errors entirely, I recommend choosing Gemma 4 31B and locking your node settings to it. By explicitly setting the model to match your environment, you can maintain a seamless automation pipeline even on the free plan.
🔗 Original Post :
제미나이 API 무료플랜 추천 모델 2종, n8n 자동화 429 에러 해결하기
https://blog.naver.com/PostView.naver?blogId=k5kun&logNo=224299456322