DeepSeek models

DeepSeek is a Chinese AI lab that publishes large mixture-of-experts models with open weights and runs a low-cost first-party API. Its V4 generation (April 2026) introduced 1M-token context, and V4.1-Flash (September 2026) moved to a new encoder-decoder architecture with image input. The API uses OpenAI- and Anthropic-compatible formats and charges half price outside peak hours.

2 models trackedData checked October 10, 2026

Lineup

Model	Status	Weights	Tier	Context	Input / 1M	Output / 1M	Released
DeepSeek-V4-ProDeepSeek V4	Current	Open weights	Frontier	1M	$1.32	$3.96	Apr 24, 2026
DeepSeek-V4.1-FlashDeepSeek V4	Current	Open weights	Advanced	1M	$0.30	$1.20	Sep 10, 2026

The lowest-priced current model is DeepSeek-V4.1-Flash at $0.30 input and $1.20 output per million tokens. The most capable tier in the lineup is frontier, led by DeepSeek-V4-Pro. The longest context window is 1M tokens (DeepSeek-V4-Pro).

Frequently asked questions

What models does DeepSeek offer?

As of October 10, 2026, we track 2 DeepSeek models: DeepSeek-V4-Pro, DeepSeek-V4.1-Flash. Status, prices and context windows for each are in the table on this page.

How do I access DeepSeek models?

Through DeepSeek's developer platform (api-docs.deepseek.com), in the DeepSeek app, and for many models through cloud marketplaces and inference providers listed on each model page.

Does DeepSeek release open-weight models?

Recent models, including DeepSeek-V4-Pro and V4.1-Flash, are released under the MIT license on Hugging Face.

Other providers

OpenAI Anthropic Google xAI Amazon (AWS)Meta Perplexity Mistral AI Alibaba (Qwen)Microsoft (Phi)NVIDIA Moonshot AI Z.ai (Zhipu AI)MiniMax IBM Ai2 (Allen Institute for AI)

Still choosing a model?

Send us your use case and constraints. We'll come back with a shortlist and an evaluation plan you can run on your own data.

Free and vendor-neutral: no model provider pays for placement.
Your details go only to partners you agree to.