Compare AI models

Specifications
Parameters
1.6T1.05T
Context window
1M1M
API pricing
Input price
$0.66$0.70
Output price
$1.98$2.10
Cached input price
$0.022$0.07
Cheapest input
$0.2066Relace$0.70Mistral
Cheapest output
$0.689Baidu$2.10Mistral
$80.39Baidu · fp8$245Mistral list price
Benchmarks
BullshitBench v2
35%80%
Terminal-Bench 4.0
6.67%41%
BrowseComp
83.4%84.5%
AutomationBench
31.8%59.9%
Overview
CompanyDeepSeekMistral
Release dateAug 13 2026Oct 6 2026
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

Mistral Large 4 leads DeepSeek-V4-Pro-0813 on 4 of the 4 benchmarks they both report (BullshitBench v2, Terminal-Bench 4.0, BrowseComp, AutomationBench). DeepSeek-V4-Pro-0813 is cheaper on both input and output: $0.66 vs $0.70 per million input tokens, and $1.98 vs $2.10 per million output tokens. Figures are base-tier rates. DeepSeek-V4-Pro-0813 shipped 54 days before Mistral Large 4, so benchmark comparisons should account for the intervening progress.

DeepSeek-V4-Pro-0813 has 1.6T parameters, while Mistral Large 4 has 1.05T. Context windows are 1M (DeepSeek-V4-Pro-0813) vs 1M (Mistral Large 4).

On BullshitBench v2, Mistral Large 4 leads at 80% vs DeepSeek-V4-Pro-0813 at 35%. On Terminal-Bench 4.0, Mistral Large 4 leads at 41% vs DeepSeek-V4-Pro-0813 at 6.67%. On BrowseComp, Mistral Large 4 leads at 84.5% vs DeepSeek-V4-Pro-0813 at 83.4%. On AutomationBench, Mistral Large 4 leads at 59.9% vs DeepSeek-V4-Pro-0813 at 31.8%.