Compare models

Pick up to 4models. Specs render side-by-side. Share the URL — it's stateless.

Selected

Llama 4 Maverick×

Mistral Large 2×

DeepSeek-V3.1×

DeepSeek-V4-Pro×Clear all

	Llama 4 Maverick Meta	Mistral Large 2 Mistral	DeepSeek-V3.1 DeepSeek	DeepSeek-V4-Pro DeepSeek
Vendor	Meta	Mistral	DeepSeek	DeepSeek
Family	Llama	Mistral	DeepSeek	DeepSeek
Release date	2025-04-05	2024-07-24	2025-08-21	2026-04-22
Context window	1,000,000 tokens	128,000 tokens	128,000 tokens	1,000,000 tokens
Parameters	400B	123B	671B	1.6T (49B active)
Modality	text, vision	text	text	text
License	Llama 4 Community License	Mistral Research License	MIT	MIT
Source	open weights	open weights	open weights	open weights
Description	Mixture-of-experts open-weight model from Meta.	—	Large MoE open-weight model. Predecessor to DeepSeek-V4.	DeepSeek's flagship open-weight MoE. 1.6T parameters with 49B activated, 1M-token context, and a hybrid attention scheme (CSA + HCA) that delivers long-context inference at ~27% of V3.2's FLOPs.
Links	weights →	weights →	weights →	weights →