Remix.run Logo
▲ sroussey 7 hours ago

Deepmind creates these, Google Cloud productizes them.

If Google Cloud was in charge, Gemini 5 would be public, and they wouldn't be messing with Gemini 4 nor 1M context when they have 10M internal. C'est la vie.

▲bushbaba 7 hours ago | parent | next [-]

10M would be huge. I have soo many use cases that could leverage it. Pls ship it google if your listening.

▲mixtureoftakes 7 hours ago | parent [-]

google, pls. Please

▲igleria 7 hours ago | parent | prev | next [-]

this is like a rumor of a rumor at this point

▲sroussey 6 hours ago | parent [-]

Or a drink with deep mind employee…

▲guelo 7 hours ago | parent | prev | next [-]

A 10M window would require dozens of GPUs (or TPUs) dedicated to each query. That is something that Google could afford internally but wouldn't be able to scale to an external product.

▲anticensor 5 hours ago | parent [-]

More like one huge waferscaler such as Cerebras.

▲tex0 7 hours ago | parent | prev [-]

Show me the 10M internal, please.