Why it matters
Model eval is where token spend decisions live. Understanding which model is actually better for your coding tasks directly informs routing choices that cut waste — this dinner gets at the measurement layer beneath every cost optimization.
The tokenmaxxing angle
Evaluation bottlenecks determine which model you route to and how much you trust its output. Poor eval = overspending on the wrong model or tier. The infrastructure and methodology discussed here is foundational to any intelligent model-routing strategy.
From the organizers
Speakers include Mark Hoffmann (Staff ML Engineer, Meta) and Gabe Greenberg (Founder & CEO, G2i); the framing is that 'evaluation is becoming the bottleneck' as coding capabilities advance.