DeepSeek

Compare the 6 models in DeepSeek — independently verified, re-verified on a rolling schedule, graded on capability, pricing and compliance.

What is in DeepSeek

HokAI tracks 6 active tools in DeepSeek. 2 of them (33%) publish a free tier you can test before paying.

DeepSeek — Models

  • DeepSeek V4 — DeepSeek V4-Pro reached GA in August 2026 under MIT, matching frontier models on agentic coding. Open-source MoE, 1M-token context, peak/off-peak API pricing.
  • DeepSeek-V4 Flash — DeepSeek-V4 Flash: 284B-param MoE with 13B active (April 2026), 1M context, 79% SWE-bench, $0.14/M input. Open-source MIT license.
  • DeepSeek-V4-Flash-0731 — DeepSeek-V4-Flash-0731 (2026) is a 284B MoE model with 13B active params, 1M-token context, and 88.1 GPQA Diamond.
  • DeepSeek-V4-Flash-Vision-Exp — DeepSeek-V4-Flash-Vision-Exp (2026) is an experimental multimodal fork of V4-Flash, adding image input at unchanged per-token pricing and a 1M-token context window.
  • DeepSeek-V4-Pro — DeepSeek V4 Pro: 1.6T-param open-source MoE (April 2026), 80.6% SWE-bench Verified with 1M token context under MIT license. $1.74/$3.48 per 1M tokens.
  • DeepSeek-V4-Pro-0813 — DeepSeek-V4-Pro-0813 is DeepSeek's flagship 1.6T-parameter mixture-of-experts model, generally available since August 2026 with MIT-licensed open weights.

Frequently asked questions

How many models are there in DeepSeek?

HokAI currently lists 6 active entries in DeepSeek. Each is verified against the vendor's own pages and re-checked as those pages change, so the count moves as the category does.

Which models are listed under DeepSeek?

DeepSeek V4, DeepSeek-V4 Flash, DeepSeek-V4-Flash-0731, DeepSeek-V4-Flash-Vision-Exp and DeepSeek-V4-Pro, among the 6 entries in this section. Each links through to a full profile covering what it does, how it is priced where that is published, and how it compares with the alternatives in the same category.

Are there free DeepSeek options?

Yes — 2 of the 6 listings in DeepSeek publish a free tier. Free tiers here usually cap usage volume or hold back the higher-end features rather than expiring, so they suit evaluation before committing.