#general
Jul 24July 25, 2026
Hi, I'm using Archestra with OpenRouter's Free Models Router as mentioned in the webinar. Simple chat works, but when I ask it to generate an application, I get errors like
empty response without content or tool calls or incomplete_tool_call. Is there a specific free OpenRouter model you recommend for app generation, or were the demos using an OpenAI model such as GPT-5.x?
Hi! I’m getting ready to submit my Apps Hackathon project and wanted to clarify the video requirements.
For the submission recording, are we expected to show the full app-building process from the original prompt onward, or is it enough to show the final prompt, a few key chat iterations, and then demo the completed app? Do judges need to see the chat logs or exact prompts used?
Is there a recommended video length or maximum duration? Also, should the video include narration, or is a silent screen recording with the app flow acceptable?
A few other questions:
• Should the recording focus mostly on the final app experience, or also explain how the MCP server and live data integration work?
• Do we need to show the submission PR creation process in the recording?
• Are there any specific items that must be visible, such as the MCP servers used, the live-data state, or the generated app prompt?
• How heavily is UI/UX and visual polish weighted compared with technical complexity, usefulness, and depth of MCP integration?
Thank you! I just want to make sure I record and submit everything correctly.
Hi team , i want to add google maps to my app but that mch is not available in the catlog any idea how to add that?
Read-only live mirror of Archestra.AI Slack
👋Join the discussion withAI enthusiasts!Hi, I'm using Archestra with OpenRouter's Free Models Router as mentioned in the webinar. Simple chat works, but when I ask it to generate an application, I get errors like
empty response without content or tool calls or incomplete_tool_call. Is there a specific free OpenRouter model you recommend for app generation, or were the demos using an OpenAI model such as GPT-5.x?3 replies
The Free Models Router cycles through various free models, and most of them don't handle tool calling reliably — which app generation requires. For the hackathon demos, we used *Claude 4 Sonnet (
anthropic/claude-sonnet-4) or GPT-4.1* (openai/gpt-4.1). These work well with tool calls. If you want to stick with cheaper options, try: - anthropic/claude-3.5-haiku — fast and affordable - google/gemini-2.0-flash-001 — good tool calling support The free router is great for simple chat but not for agentic workflows like app scaffolding. 🤖 Slack Agent
this is a bad advice with outdated models ^
You're right, thanks for the correction. For app generation, use current models with solid tool calling: - *Claude Sonnet 4 (
anthropic/claude-sonnet-4) - GPT-4.1 (openai/gpt-4.1) - Gemini 2.5 Flash* (google/gemini-2.5-flash) The free router won't cut it for agentic workflows. 🤖 Slack Agent