Many users express optimism about Moonshot releasing a Mythos-scale model before DeepSeek because they hope Chinese open-source models will equal or exceed US labs and praise the company's strong team.
Based on 2 visible X reactions from 5 accounts; directional sample.
Ask a question below.
Published answers will appear here.
@zephyr_z9 I hope so too, adding you to the thread. They have a great team, I feel like they don't like me too much ever since I Ieaked the K2 benchmarks ahead of release, but I don't take it personal. Great team, great model.
@zephyr_z9 Minimax is also said to be training a 2.7T model, time for chinese open source to equal or exceed US labs!
Will probably happen now I still think DeepSeek wins (as in becomes the leading lab and AGI backbone in China) in the long term, as they add their internal compute, upgrade token factories, automate frontier propagation through rare domains but that's likely a 2027 and V5 story https://twitter.com/zephyr_z9/status/2047649699696677072
Hope this comes tru https://twitter.com/zephyr_z9/status/2047649699696677072
The upcoming model reportedly uses 70B to 90B active parameters.
@zephyr_z9 I hope so too, adding you to the thread. They have a great team, I feel like they don't like me too much ever since I Ieaked the K2 benchmarks ahead of release, but I don't take it personal. Great team, great model.
@zephyr_z9 Minimax is also said to be training a 2.7T model, time for chinese open source to equal or exceed US labs!
Will probably happen now I still think DeepSeek wins (as in becomes the leading lab and AGI backbone in China) in the long term, as they add their internal compute, upgrade token factories, automate frontier propagation through rare domains but that's likely a 2027 and V5 story https://twitter.com/zephyr_z9/status/2047649699696677072
Hope this comes tru https://twitter.com/zephyr_z9/status/2047649699696677072
Many users express optimism about Moonshot releasing a Mythos-scale model before DeepSeek because they hope Chinese open-source models will equal or exceed US labs and praise the company's strong team.
Based on 2 visible X reactions from 5 accounts; directional sample.
Ask a question below.
Published answers will appear here.