deepseek-v4.1IMAGE

DeepSeek V4.1 Flash scores 80.3 for $0.83 on the 23-model arnavbench

On the author's custom benchmark, Opus 5.5 tops at 88.9 for $8.89 a run, while DeepSeek V4.1 Flash gets 80.3 for just $0.83, a standout price/perf result.

Arnav Prabhu

@ArnavPrabhuArchived 11 days agoOriginally posted 2026.10.18

DeepSeek V4.1 Flash scores 80.3 for $0.83 on the 23-model arnavbench (image 1 of 3)

DeepSeek V4.1 Flash scores 80.3 for $0.83 on the 23-model arnavbench (image 2 of 3)

DeepSeek V4.1 Flash scores 80.3 for $0.83 on the 23-model arnavbench (image 3 of 3)

Arnav Prabhu @ArnavPrabhuOriginally posted 2026.10.18

Open original on X

arnavbench now has 23 models Opus 5.5 still on top at 88.9 ($8.89 a run) But the price/perf is WILD: Deepseek v4.1 flash: 80.3 for $0.83, beats sonnet 5.5 Qwen3.8 27b running locally on my mac: 76.9 for ~5 cents of electricity DISCLAIMER: tuned to my work (rust, LLM inference tuning, finance ops). your results may vary. make your own bench, it's worth it

x.com/arnavprabhu/status/2111541097298329602Archived 2026.10.18 16:29