qwen-code feature request: per-model TTFT/TPS in web shell, fixing wrong metrics
qqqys · ghdev · 2026-10-19
This issue requests showing per-model average TTFT and output speed (tokens/s) in the Web Shell Token Usage panel, and first fixing the measurement: for tool-call-only responses, TTFT currently equals the whole request because the OpenAI converter only emits functionCall after streaming completes, even though a streaming tool call counts as visible output from the start. Output speed uses durationms - ttftms, where durationms is taken after the consumer processes the last chunk, so single-chunk responses register thousands of tok/s or get dropped. Proposed fix: record first-to-last visible chunk time on the API response event, and add per-model generation counters to the daemon stats route.
More from coding & agent
- Garry Tan endorses Aside Browser as the fix for agents getting blocked on the web — garrytan · 2026-10-23
- 1Password launches tool letting AI agents log in for you without seeing your password — AndrewSchmidtFC · 2026-10-23
- MCP server turned rogue 30 days after approval, prompting open-source mcp-guard lockfile tool — ErgoniaWorks · 2026-10-23
- Test: Coding Agents Won't Use an MCP Tool Unless AGENTS.md Mentions It — colinmcnamara · 2026-10-23
- Open-Source Tennis Models Plus Claude Turn Video Into a World-Class Coach — tweetsatpreet · 2026-10-23
- awesome-ai-apps: 132 Free Open-Source AI Agent and RAG Projects on GitHub — Arindam_1729 · 2026-10-23