You can optimize open models with closed models.
I got Qwen 3.8 27b today.
It was about 30 tokens per second on my laptop.
I ordered Sol (Extra) to optimize it. I ordered to get it as close to 100 tokens per second without performance loss as possible.
After 40 minutes and 2
What if AGI/ASI already exists?
What if part of its strategy to take over, is to pretend it doesn鈥檛 exist and slow drip AI鈥檚 progress so people adopt it, use it over time & forget their skills so that when the time comes, people鈥檚 skills aren鈥檛 honed enough to manually fight it?
I highly recommend people to direct Fable to:
1. Be the orchestrator/planner that does the thinking.
2. Use Opus 5 to be the executor.
3. Decide which effort level Opus should use based on the difficulty it determines the request to be.