What will be true of DeepSeek's r2 model?
16
1.3kṀ1318Mar 17
86%
Some version of it will score better on GPQA than o1-preview (73% pass@1)
62%
Its base model will be DeepSeek-V3
48%
It will score >= 78% on LiveBench when first added (R1 = 71.5% on March 12)
45%
DeepSeek will report it scores >= 65% on SWE-Bench Verified
45%
It will have a variant with more parameters than DeepSeek-r1 (not necessarily active)
45%
Manifold thinks it is better than expected in a poll asking if it is better or worse than expected
42%
It will support a longer context length than DeepSeek-r1
10%
Reaches Highest Arena Score on Chatbot Arena
everything N/As if a model called R2 created by DeepSeek is not released in 2025
This question is managed and resolved by Manifold.
Get
1,000 to start trading!
1,000Related questions
Related questions
When will DeepSeek release R2?
Will DeepSeek's next reasoning model be called R3?
4% chance
Will DeepSeek's next reasoning model be open-sourced?
91% chance
Will DeepSeek R2 be open source?
83% chance
When will DeepSeek release V4?
Will I believe in 1 year that DeepSeek R1 was substantially trained via distillation of a US model?
64% chance
Will DeepSeek be banned in US before 2027?
25% chance
How much did DeepSeek-V3 cost to train?
Will ANY DeepSeek model cause or materially enable a catastrophic risk by 2027?
12% chance
Did DeepSeek receive unannounced assistance from OpenAI in the creation of their v3 model?
8% chance