Performance Benchmarks for devstral-2

Analyze how devstral-2 stacks up against other frontier models in terms of cost-to-performance and raw intelligence (LMArena Elo).

2026 Performance Summary

🚀 Intelligence (Elo): 1194
💰 API Cost (Input): $4.4e-7 / 1M tokens
⚡ API Cost (Output): $0.0000022 / 1M tokens
⏱️ Real-world Throughput: 27 tokens/s

Analysis: While devstral-2 is a capable model, it is currently outperformed in efficiency by muse-spark-1.1. muse-spark-1.1 sits higher on the Pareto line, offering approximately 297 points more intelligence (Elo) for a comparable or lower API cost.

Highly Recommended Alternatives

Frontier Model	LMArena Elo	API Cost (1M)	Throughput
muse-spark-1.1	1491	$0.00000125	268
gemini-3-flash	1473	$5e-7	81
mimo-v2.5-pro	1468	$4.35e-7	97
hy3	1457	$1.32e-7	72
gemma-4-31b	1451	$1e-7	213

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