As of , you should be using...

BEST MODEL
Fable 5
WORKHORSES
Opus 5 Medium · GPT 5.6 Sol High
Have Fable 5 orchestrate them
I'm on a budget.
$100 Claude // $20 OpenAI is still the move.
Google? Meta Muse Spark?
Extremely Mid
What about open models?
GLM 5.2 and Kimi K3 are good. They're slow and think too much. But good.
IQ/$ maxxing?.
Grok 4.5 · DeepSeek V4 Flash · GPT 5.6 Luna Max are pareto optimal
What if I'm curing cancer or proving theorems?
GPT Pro series is still the king of this kind of thing. Might not be many theorems left to prove soon.
What about Cursor/OMP/OpenCode/3P Harness?
Don't waste your time
Is OpenClaw still a thing?
Not really. Probably just use Hermes.
What about running models locally?
Fun thing to screw around with for a bit.
What about [benchmark]?
Vibe check is the only benchmark that matters.
@eatpraydiehard · Get Updates