Guys, this is why Claude Mythos is dangerous.
Claude Opus 4.6 just proved how to turn those 5-minute high-leverage windows on Polymarket into money printing machines. Starting at $1,000, same market, 48 hours. A Claude robot did 14,216 knives, +1322%. The other goes straight to zero. The difference is not whether the guess is accurate or not, but who can write the discipline into the code.
Grok 4.6 High vs Opus 5 High vs Fable 5 High vs GPT 5.6 sol High
Why pay $22 and wait 53 minutes for opus 5 when grok 4.6 can run the exact same prompt in 27 minutes for $1.67?
I ran a simple experiment and gave 2 top models the exact same prompt. results: grok 4.6: 27 min, $1.67 opus 5 max: 53 min, $22 that makes grok roughly 2x faster
Tencent Hunyuan Hy4 preview released: directly competing with Sol and Opus 5
You have reached the end of the archive
All of Claude Opus 5