[BUG] LLM Request Failed - 429 rate limit

[xalgord]

I tried using the Gemini 2.5 Pro and Gemini 2.5 Flash models but got rate limited so early. Is there any feature to use multiple API keys at once?

[Armankb2]

Hi @xalgord 👋,

I’d love to help with this issue!

From the error logs, it looks like the LLM request is hitting a 429 rate-limit when using the Gemini 2.5 / 2.5 Flash models.
This usually happens because the current API key quota is exhausted or requests are being sent too quickly without proper throttling.

🧩 Possible Fix Ideas:

• Implement a retry/backoff mechanism for 429 responses.
• Add support for rotating multiple API keys (as you mentioned) — for example, cycling through keys stored in .env variables.
• Add a configurable request delay or queue system in Strix to avoid hitting provider limits.
• Log the number of requests sent per minute to help monitor usage.

I don’t have my laptop right now, but I’d be happy to work on this fix collaboratively once I’m back on my setup.
Please assign or tag me when it’s open for contributors 🙌

Thanks!
— @Armankb2

[xalgord]

also which model would be better for web application testing, chatgpt 5 or gemini 2.5 pro?

[fortinixeyas]

hello i tried to use the gemini flash 2.5 but i have this problem :
export STRIX_LLM="google/gemini-2.5-flash"

[clgp-aint-cool]

@fortinixeyas it should be
export STRIX_LLM="gemini/gemini-2.5-flash"

You can see other models at https://docs.litellm.ai/docs/

[andy-wei]

hello i tried to use the gemini flash 2.5 but i have this problem : export STRIX_LLM="google/gemini-2.5-flash"

You can change it to export STRIX_LLM="gemini/gemini-2.5-flash" and the problem should be fixed.

[xalgord]

I switched to Paid GPT-5 API, thanks.

[khuluqilkarim]

hello i tried to use the gemini flash 2.5 but i have this problem : export STRIX_LLM="google/gemini-2.5-flash"

You can change it to export STRIX_LLM="gemini/gemini-2.5-flash" and the problem should be fixed.

i got this issue , can u help me

[xalgord]

It should be gemini/gemini-2.5-flash and not google/gemini-2.5-flash

[TahaDataAlchemy]

Strix works perfectly with GPT-4o, Gemini 2.5 Pro, and DeepSeek R1, but throws repeated LiteLLM errors with other models like (Mistral 8B, Qwen 2.5 72B) even though API configuration is identical and valid.

Why do some models work while others don't?
Is this a LiteLLM compatibility issue or a model capability issue?

using this command
strix -n --target https://

[mrwiora]

I included strix into my github workflows on my project.
I have a paid Claude Platform Subscription.

Unfortunately I also got

❌ Penetration test failed: LLM request failed: Rate limit exceeded

╭────────────────────── 🔍 Live Penetration Test Status ───────────────────────╮
│                                                                              │
│  🦉 Running penetration test...                                              │
│                                                                              │
│  🧠 Model: ***                                       │
│  🔍 Vulnerabilities: 0                                                       │
│  🤖 Agents: 7 • 🛠️ Tools: 87                                                  │
│  📥 Input: 2.1M • ⚡ Cached: 1.6M                                            │
│  📤 Output: 23.7K • 💰 Cost: $2.6737                                         │
│                                                                              │
╰──────────────────────────────────────────────────────────────────────────────╯
Error: Process completed with exit code 1.

Reference: https://github.com/mrwiora/tpm2-kira/actions/runs/20994975512/job/60351631624

Unfortunately nothing has been uploaded - maybe it would be possible - to in any case upload what already has been produced.

@Armankb2 shall I open a new issue for that? Created the answers here as it seems very related, but realized it is closed.

[mrwiora]

Requested Claude to support on this, by requesting higher Rate Limits.

Actually I see the hitted limits: