- Google upgraded its AI offering with Gemini Spark, a new agent that responds faster and with fewer errors than comparable services.
- In PCMag’s hands, Spark answered 92% of test prompts correctly and returned results in roughly 0.9 seconds per query, whereas the next‑best agent averaged 1.3 seconds and hit correct answers on 78% of prompts (PCMag). The test also noted Spark’s lower rate of “hallucinations” – about 4% versus 11% for the competitor.
- The speed and reliability matter because developers can now build tighter conversational loops without the latency that often forces workarounds. However, the review warns that Spark’s value proposition is murky; it duplicates much of Gemini’s core functionality, leaving customers to wonder why they need both models.
- Until Google clarifies the niche Spark fills, the agent may remain a nice‑to‑have add‑on rather than a must‑have component of its AI suite.