StorySoftware
Google DeepMind announces Gemini 1.0
Google introduced Gemini 1.0, a model built from the start to handle text, code, audio, images and video together, in three sizes: Ultra, Pro and Nano (for on-device use). Google said Gemini Ultra beat the previous best results on 30 of 32 common academic benchmarks and was the first model to outperform human experts on the MMLU knowledge test.
- Natively multimodal, processing text, code, audio, images and video simultaneously from the ground up.
- First model to outperform human experts on MMLU benchmark with 90.0% accuracy.
- Exceeds state-of-the-art on 30 of 32 widely-used LLM research benchmarks.
- Three sizes: Ultra (complex tasks), Pro (scaling), Nano (on-device); Gemini Pro launched December 13.