LLM ReferenceLLM Reference
All comparisons

DeepSeek V3 vs o3

Side-by-side comparison of specifications, capabilities, and pricing.

Released2024-12-262025-03-31
Context window64k128K
Parameters671B
Architecturemixture of expertsdecoder only
LicenseOpen SourceUnknown
Knowledge cutoff2024-04

Capabilities

Vision
Multimodal
Reasoning
Function calling
Tool use
Structured Outputs
Code execution

Availability

Providers

Benchmarks

HellaSwag95.7—HumanEval85.596.7Massive Multitask Language Understanding88.5—LiveCodeBench49.679.1Aider Polyglot48.481.3BigCodeBench50.0—Chatbot Arena1302.01412.0MMLU PRO75.9—SWE-bench Verified—71.7Google-Proof Q&A—87.7Massive Multi-discipline Multimodal Understanding—82.9