Performance Benchmarks for hy3

Analyze how hy3 stacks up against other frontier models in terms of cost-to-performance and raw intelligence (LMArena Elo).

2026 Performance Summary

🚀 Intelligence (Elo): 1453
💰 API Cost (Input): $1.32e-7 / 1M tokens
⚡ API Cost (Output): $5.28e-7 / 1M tokens
⏱️ Real-world Throughput: 57 tokens/s

Analysis: While hy3 is a capable model, it is currently outperformed in efficiency by DeepSeek-V4.1-Flash-20261017 (Max). DeepSeek-V4.1-Flash-20261017 (Max) sits higher on the Pareto line, offering approximately 211 points more intelligence (Elo) for a comparable or lower API cost.

Highly Recommended Alternatives

Frontier Model	LMArena Elo	API Cost (1M)	Throughput
DeepSeek-V4.1-Flash-20261017 (Max)	1664	$3e-7	131
Claude Haiku 5.5 (High)	1586	$1e-7	166
glm-5.3-max	1479	$7.9e-8	77
deepseek-v4.1-flash-max	1475	$3.52e-8	66
gpt-oss-20b	1317	$1.8e-8	26

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