Gemini Live
- Strength
- Second-highest repetition score at 4.78/5 across 174 scored calls.
- What can be improved
- One rerun had no latency evidence and scored 0/5 for both task outcome and tool accuracy.
- reasoning: high
- hybrid VAD
Gemini Live is ahead on tool call accuracy, interruption handling and response time, and Vapi on repeatable reliability, task completion and infrastructure reliability.
Last updated Methodology by Luis Ojeda
Pick Gemini Live for accurate tool calls (0.38 higher), handling interruptions (0.24 higher) and fast replies (0.03s faster). Pick Vapi for reliability across repeated runs (29.27 points higher), completing the caller's task (9.76 points higher) and calls that connect and stay up (10.57 points higher).
Bars share one scale across all 8 platforms. The place next to each figure is its rank in the field.
Share of the 82 scenarios that passed on all three runs.
Vapi, 29.27 points higher
Calls where the expected outcome was fully reached.
Vapi, 9.76 points higher
Calls with no connection, audio or platform failure.
Vapi, 10.57 points higher
Mean score for calling the right tool with the right arguments.
Gemini Live, 0.38 higher
Mean score for how clear and natural the agent sounds.
Not comparable
Mean score for yielding and recovering when the caller cuts in.
Gemini Live, 0.24 higher
Mean time for the agent to start replying after the caller stops.
Gemini Live, 0.03s faster
246 calls per platform: 82 scenarios, 3 runs each.
Vapi is more reliable, by 29.27 points. Gemini Live passed 30.49% of the 82 scenarios on all 3 runs and Vapi passed 59.76%. A scenario only counts when every run of it passed.
Gemini Live starts replying sooner, 0.03s faster. Mean response time is 3.05s for Gemini Live and 3.08s for Vapi.
Gemini Live reached the expected outcome on 87.80% of calls and Vapi on 97.56%. Vapi is ahead on task completion.
Both ran the same Appointment and Medicare agents through the same 82 caller scenarios, 3 times each, with the same evaluators and mock tools. Only the platform and its speech and model components changed.