Performance Benchmarks for deepseek-v4-flash-thinking

Analyze how deepseek-v4-flash-thinking stacks up against other frontier models in terms of cost-to-performance and raw intelligence (LMArena Elo).

2026 Performance Summary

🚀 Intelligence (Elo): 1439
💰 API Cost (Input): $1.12e-7 / 1M tokens
⚡ API Cost (Output): $2.24e-7 / 1M tokens
⏱️ Real-world Throughput: 55 tokens/s

Analysis: deepseek-v4-flash-thinking is currently positioned on the Pareto Frontier. This means it offers a unique and optimal balance between intelligence (Elo: 1439) and cost ($1.12e-7/1M tokens), making it a top-tier choice for its price bracket.

Highly Recommended Alternatives

Frontier Model	LMArena Elo	API Cost (1M)	Throughput
gemini-3.1-pro-preview	1488	$0.000002	79
gpt-5.5	1477	$0.00000125	31
gemini-3-flash	1473	$5e-7	76
deepseek-v4-pro-thinking	1461	$4.35e-7	27
gemma-4-31b	1451	$1.2e-7	13

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