news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

it home reported on september 25 that the technology media the decoder published a blog post yesterday (september 24), reporting that google upgraded its gemini 1.5 ai model and launched Gemini-1.5-Pro-002 andGemini-1.5-Flash-002compared with previous versions, it has lower cost, stronger performance and faster response.

lower costs

google has lowered token input and output fees by up to 50% for gemini-1.5-pro-002 and gemini-1.5-flash-002, increased rate limits for both models, and reduced latency.

the new pricing takes effect on october 1, 2024.

better performance

it home quoted the press release and attached the performance of the new model as follows:

on the more challenging mmlu-pro benchmark, the model’s performance improves by about 7%.

math performance has been significantly improved by 20% in the math and hiddenmath benchmarks.

improvements were also seen on vision- and code-related tasks, with gains of 2-7% in the visual understanding and python code generation evaluations.

google claims that these models now provide more helpful responses while maintaining content safety standards. based on developer feedback, the company has improved the output style of the models, aiming for more precise and cost-effective use.

other improvements

google also upgraded the gemini 1.5 experimental model released in august and launched Gemini-1.5-Flash-8B-Exp-0924 the upgraded version further enhances text and multimodal applications.

the new gemini models are available through google ai studio, the gemini api, and vertex ai (for google cloud customers). a chat-optimized version of gemini 1.5 pro-002 for gemini advanced users is coming soon.