gemma-4-31b vs grok-4.20-beta-0309-reasoning

Comparing these models based on real-world throughput and community-driven performance benchmarks.

Direct Technical Comparison

Frontier Model	LMArena Elo	API Cost (1M)	Throughput
gemma-4-31b	1452	$1.4e-7	9
grok-4.20-beta-0309-reasoning	1481	$0.000002	22

*These models represent the Pareto Frontier (optimal cost-to-performance).*

Conclusion: For developers looking for higher intelligence, grok-4.20-beta-0309-reasoning currently leads with an Elo of 1481.