Skip to content
llm-speed

DeepSeek-Coder-V2-Lite-Instruct vs Codestral-22B-v0.1

Side-by-side decode tok/s, prefill tok/s, and TTFT for DeepSeek-Coder-V2-Lite-Instruct and Codestral-22B-v0.1, sourced from community-submitted runs of the llm-speed suite. Every number on this page links back to the run it came from.

model
DeepSeek-Coder-V2-Lite-Instruct
DeepSeek · 16B-A2.4B
View DeepSeek-Coder-V2-Lite-Instruct page →
model
Codestral-22B-v0.1
Mistral · 22B
View Codestral-22B-v0.1 page →

Verdict

On the M3 Ultra (60-core GPU), DeepSeek-Coder-V2-Lite-Instruct decodes at 168 tok/s versus 47 tok/s for Codestral-22B-v0.1, 3.5× faster. That is the only hardware measured on both so far; the row below links to both source runs. Submit another to widen the comparison.

Hardware with data on both models

HardwareDeepSeek-Coder-V2-Lite-Instruct decodeCodestral-22B-v0.1 decodeΔSource runs
M3 Ultra (60-core GPU)168.3tok/s47.49tok/s+120.9r_l_v1-zq_qaz · r_79dvtag5fd_

See also: DeepSeek-Coder-V2-Lite-Instruct benchmarks · Codestral-22B-v0.1 benchmarks · All hardware · All models · Methodology