Compare
Grok 4 vs DeepSeek V4
xAI vs DeepSeek. Specs, benchmarks, and real per-task cost — all in one page.
Verdict
Grok 4 leads on LMSYS ELO (1358 vs 1334). DeepSeek V4 is ~42.9× cheaper on a 3:1 input:output blend. DeepSeek V4 is open-weights, the other is proprietary.
xAI
Grok 4xAI’s latest. Real-time X search integration, strong on current events and meme-literate tasks.
Live web/X accessPersonalityFast inference on Colossus
DeepSeek
DeepSeek V4The shock-the-market moment of 2026. Frontier coding quality at 1/50th the price of Opus.
Unbeatable $/qualityOpen weightsGreat at code
Pricing
Input / 1M$5.00$0.14
Output / 1M$15.00$0.28
Context256K128K
Max output32K16K
LicenseProprietaryOpen weights
Released2026-03-012026-02-20
Benchmarks
LMSYS ELO1358.01334.0
MMLU Pro86.285.4
HumanEval88.491.5
SWE-Bench—54.1
MATH86.588.7
IFEval85.085.8
Per-task cost
Summarize a 1-hour meeting transcript$0.082$0.0022
Review a 500-line pull request$0.070$0.0017
Answer a customer support ticket$0.027$0.0006
Extract structured data from a resume$0.040$0.0010
Debug a stack trace with context$0.060$0.0014
Per-call cost using published token counts for each task. Real-world prompts vary.