Playwright not executed when using openrouter models like meta-llama/llama-4-maverick:free gives error 400 Input required: specify "prompt" or "messages"

[roman-doubrava]

What happened?

Tool use like playwrigth not executed when using openrouter models like meta-llama/llama-4-maverick:free
An Eror message occurs: Something went wrong. Here's the specific error message we encountered: An error occurred while processing the request: 400 Input required: specify "prompt" or "messages"

The same prompt works with anthropic Claude 3.7+ without any issues.

Version Information

a2f330e

Steps to Reproduce

1. configure OpenRouter and select Model meta-llama/llama-4-maverick:free
2. configure tool use filesystem + playwright
3. Prompt: Webscrape the website XXX and tell me about their products, make screenshots of each product side and create a HTML Summary file with the screenshots and the product descriptions. Use tool playwright to scrape the website."
4. Error: Something went wrong. Here's the specific error message we encountered: An error occurred while processing the request: 400 Input required: specify "prompt" or "messages"
5. When deselecting all tools use the model delivers an standard answer

What browsers are you seeing the problem on?

Chrome

Relevant log output

2025-06-10T08:00:06.990Z debug: [saveConvo] api/app/clients/BaseClient.js - saveMessageToDatabase #saveConvo
2025-06-10T08:00:07.542Z error: [api/server/controllers/agents/client.js #sendCompletion] Operation aborted 400 Input required: specify "prompt" or "messages"
2025-06-10T08:00:07.542Z error: [api/server/controllers/agents/client.js #sendCompletion] Unhandled error type 400 Input required: specify "prompt" or "messages"
2025-06-10T08:00:07.551Z debug: [saveConvo] api/app/clients/BaseClient.js - saveMessageToDatabase #saveConvo
2025-06-10T08:00:07.560Z debug: [AgentController] Request closed


{"attemptNumber":1,"code":400,"error":{"code":400,"message":"Input required: specify \"prompt\" or \"messages\""},"headers":{"access-control-allow-origin":"*","cf-ray":"94d7554bc811d24b-FRA","connection":"keep-alive","content-length":"130","content-type":"application/json","date":"Tue, 10 Jun 2025 08:00:07 GMT","server":"cloudflare","vary":"Accept-Encoding","x-clerk-auth-message":"Invalid JWT form. A JWT consists of three parts separated by dots. (reason=token-invalid, token-carrier=header)","x-clerk-auth-reason":"token-invalid","x-clerk-auth-status":"signed-out"},"level":"error","message":"[api/server/controllers/agents/client.js #sendCompletion] Unhandled error type 400 Input required: specify \"prompt\" or \"messages\"","pregelTaskId":"f43c4484-236e-5408-8180-9c8e1a6d99ad","retriesLeft":6,"stack":"Error: 400 Input required: specify \"prompt\" or \"messages\"\n    at APIError.generate (/usr/local/LibreChat/node_modules/openai/error.js:45:20)\n    at CustomOpenAIClient.makeStatusError (/usr/local/LibreChat/node_modules/openai/core.js:302:33)\n    at CustomOpenAIClient.makeRequest (/usr/local/LibreChat/node_modules/openai/core.js:346:30)\n    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)\n    at async /usr/local/LibreChat/node_modules/@librechat/agents/node_modules/@langchain/openai/dist/chat_models.cjs:2339:29\n    at async RetryOperation._fn (/usr/local/LibreChat/node_modules/p-retry/index.js:50:12)","status":400}

Screenshots

No response

Code of Conduct

• I agree to follow this project's Code of Conduct

[danny-avila]

The same prompt works with anthropic Claude 3.7+ without any issues.

It's a context window issue, we don't have a default window size set for llama-4.

You may potentially get around it if you set an appropriate value for max context tokens:

[roman-doubrava]

@danny-avila thank you for the hint, we tried 4028 or 2048 but get always his error:

Something went wrong. Here's the specific error message we encountered: An error occurred while processing the request: 400 Input required: specify "prompt" or "messages"

According to the model card https://openrouter.ai/meta-llama/llama-4-maverick:free and https://openrouter.ai/meta-llama/llama-4-scout:free it should work with 4028.

Could you check at your end if you get any of the tools like playwrigth and filesystem running with llama-4-maverick:free ?

[iliyanpetrovcleverpine]

Hi @roman-doubrava,
I encountered the same issue and fixed it with the Danny's suggestion. You can see my comment #7943 (reply in thread).