Google & Meta AI: Gemini 3.8 vs. Muse Spark 1.3

Google and Meta both released new AI models on the same day. Google launched Gemini 3.8 Flash, along with a cybersecurity version. Meta introduced Muse Spark 1.3. Independent tests show a split result between the two. Meta's model excels in agentic knowledge work and scientific reasoning. Google's model, however, has an advantage in factual recall and terminal coding. Gemini 3.8 Flash is Google's third Flash release in six weeks. Its introductory rate is $0.75 per million input tokens and $3.75 per million output tokens until the end of 2026. After that, prices double. The cybersecurity variant, Gemini 3.8 Flash Cyber, scored 86.2% on a benchmark for finding vulnerabilities. It also produced 2.6 times more correct patches for Chrome vulnerabilities than larger commercial models. Access to this cyber model is limited to government and critical infrastructure operators. Meta's Muse Spark 1.3 showed roughly 20% fewer tool calls compared to its previous version. In independent testing, Muse Spark 1.3 scored higher in some areas, like the Sierra Research banking agent test. Gemini 3.8 Flash led in others, such as Terminal-Bench 2.1 and AA-Omniscience accuracy. The bottom line is that the competition in advanced AI is rapidly intensifying, with multiple major releases happening in a short timeframe.