All comparisons

DeepSeek V3 vs Llama 4 Maverick 17B Instruct FP8

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

	DeepSeek V3 DeepSeek	Llama 4 Maverick 17B Instruct FP8 AI at Meta
Released	2024-12-26	2026-03-01
Context window	—	128K
Parameters	—	17B
Architecture	mixture of experts	mixture of experts
License	Unknown	Open Source
Knowledge cutoff	—	—
Capabilities
Multimodal
Function calling
Tool use
JSON mode
Availability
Providers	DeepInfraFireworks AI PlatformDeepSeek PlatformMicrosoft Foundry	Microsoft Foundry

Benchmarks

HellaSwag95.7—HumanEval85.5—Massive Multitask Language Understanding88.5—