openrouter.ai · openai ·
moonshotai/kimi-k3.1-pro on openrouter.ai
passable7 of 7 cases passed
63/100trust score
Observed facts across five dimensions
- Protocol compatibilityopenai; legal tool call: yes; parameters: ok
- Agent readiness7/7 cases completed
- Observabilityrelay reported token usage
- Performancetotal duration 86,040 ms
- Reference deviationduration +70%; input tokens +25%; output tokens +138%
Seven test cases
| Case | Result | Latency | Baseline | Difference |
|---|---|---|---|---|
| Basic Tool I/O | Passed | 14,981 ms | 8,830 ms | +70% |
| Multi-File Synthesis | Passed | 25,168 ms | 18,750 ms | +34% |
| Model Context Synthesis | Passed | 6,159 ms | 2,509 ms | +145% |
| Single Tool Call | Passed | 10,310 ms | 4,534 ms | +127% |
| Chained Tool Calls | Passed | 13,168 ms | 6,472 ms | +103% |
| Error Retry & Recovery | Passed | 10,797 ms | 6,335 ms | +70% |
| Long-Context Recall | Passed | 3,379 ms | 1,881 ms | +80% |
Tokens and duration
| Metric | Relay | Baseline | Difference |
|---|---|---|---|
| Duration | 86,040 ms | 50,708 ms | +70% |
| Input tokens | 71,734 | 57,318 | +25% |
| Output tokens | 2,168 | 910 | +138% |
Previous tests
| Date | Score | Passed |
|---|
Test it again, with your own key
Last tested 2026-10-27 (1 run). Relays change — check how this one does right now with your own key (about a minute; the key is never stored).
Test it now →The relay host is prefilled; add the API path before starting.