AI Language Model
DeepSeek V3
Last updated:
91
AI SCORE
/ 100
Try or Access API
Official Product
Try for free in the browser
API Access
Integrate via API
Détail des scores
Top 30% Cost EfficiencyTop 37% Benchmark (MMLU)
Benchmark (MMLU)
90
Cost Efficiency
98
Arena ELO
78
DeepSeek V3 — Full Specifications
About the DeepSeek V3
DeepSeek's open-source frontier model with 671B MoE parameters (37B active). Exceptional performance at remarkably low cost.
Reviewed by VersusMatrix Editorial Team
Last updated: April 21, 2026
Methodology: AI-powered analysis of technical specifications from manufacturer data. Scores are calculated by comparing products across multiple dimensions and normalized relative to the full category database. Our editorial process is independent and not influenced by affiliate partnerships.
As-tu utilisé le DeepSeek V3 ?
Partage ton expérience pour aider les autres à prendre de meilleures décisions.
Les mieux notés AI Language Models
FAQ
Is the DeepSeek V3 worth buying in 2026?
DeepSeek's open-source frontier model with 671B MoE parameters (37B active). Exceptional performance at remarkably low cost.
