Dev, web & IT
Design, code, host and ship sites and internal tools, with an AI editor doing most of the typing.
Independent reviews. IonBin has no partnerships with the tools on this page and was not paid for these listings. Our evaluations are our own, based on public documentation and our research. Product names are trademarks of their owners and are used for identification only. How disclosure works.
Model · DeepSeek
Stack 08 Stack 02
DeepSeek models
Verified Oct 17, 2026 Review by Nov 16, 2026 2 sources
Short answer
DeepSeek offers two API models: deepseek-flash for low-cost volume work and deepseek-v4-pro for harder tasks. Both have a 1M-token context window, thinking and non-thinking modes, JSON output and tool calls, and off-peak use is half price. DeepSeek's privacy policy says personal data is stored in China, so review it before you send anything sensitive.
What it is
DeepSeek is a model family from the company of the same name. Its API docs list two current models and publish prices per million tokens, with separate rates for cache hits and misses and for peak and off-peak hours.
How it works
- Pick deepseek-flash for cheap, high-volume work and deepseek-v4-pro for heavier reasoning and coding. Both can run with or without a thinking step.
- Call the API with an OpenAI-format or Anthropic-format base URL, which makes it easy to try inside tools that already speak those formats.
- Schedule batch jobs off-peak. DeepSeek lists off-peak rates at half the peak price. Peak hours are 01:00–04:00 and 06:00–10:00 UTC on weekdays.
Specs and pricing (checked 2026-10-17). Both models have a 1M-token context and up to 384K output tokens. Flash costs $0.15 per million cache-miss input tokens and $0.60 per million output tokens off-peak, double that at peak. DeepSeek’s pricing page lists the older names deepseek-v4-flash and deepseek-v4-flash-vision-exp as retired, with requests now served by V4.1-Flash-1017.
Benefits
- Low list prices make it a candidate for bulk summarizing, tagging and extraction on non-sensitive data.
- The 1M-token context lets you send long documents in one request (see context window).
- Compatibility with existing API formats lowers switching cost (see avoiding lock-in).
Limits and risks
- Data location. DeepSeek’s privacy policy names Hangzhou DeepSeek AI Co., Ltd. as the controller. It says personal data is collected, processed and stored in the People’s Republic of China, and that data may be used to train and improve its models. Many businesses will need legal or client sign-off before sending client or personal data, and some won’t be able to use it for that at all.
- Vision is limited. Image input is listed for Flash only.
- Names change. Model names have already been retired once. Pin the model ID and re-check prices before you budget.
This page describes vendor terms as published. It isn’t legal advice; have counsel review data terms before you send client or personal data.
Checked specs
Snapshot from official sources on Oct 17, 2026. Model lineups and prices change often; this page is re-verified every 30 days.
| API models | deepseek-flash (DeepSeek-V4.1-Flash-1017); deepseek-v4-pro (DeepSeek-V4-Pro-0813) [1] |
| Context / max output | 1M tokens / 384K tokens [1] |
| Flash price per 1M tokens, cache-miss input / output (checked 2026-10-17) | $0.30 / $1.20 peak; $0.15 / $0.60 off-peak [1] |
| Pro price per 1M tokens, cache-miss input / output (checked 2026-10-17) | $1.32 / $3.96 peak; $0.66 / $1.98 off-peak [1] |
| Cache-hit input per 1M tokens | Flash $0.006 peak / $0.003 off-peak; Pro $0.044 peak / $0.022 off-peak [1] |
| Peak hours | 01:00–04:00 and 06:00–10:00 UTC, Monday to Friday, excluding Chinese public holidays [1] |
| Features | Thinking and non-thinking modes; JSON output; tool calls; OpenAI- and Anthropic-format APIs; vision on Flash only [1] |
| Retired model names (checked 2026-10-17) | deepseek-v4-flash and deepseek-v4-flash-vision-exp are listed as retired; requests are served by V4.1-Flash-1017 [1] |
| Search and Fetch APIs | $0.75 per 1,000 requests [1] |
| Data location (privacy policy) | Controller: Hangzhou DeepSeek AI Co., Ltd.; personal data collected, processed and stored in the People's Republic of China [2] |
Best-fit stacks
Business data & reporting
Pull numbers from the systems you already use into one place an agent can analyze and explain.
Agent compatibility
| API access | Yes | Official docs ↗ |
Levels come from each tool's recorded, sourced compatibility. How the levels work.
Related entries
Sources
- Models & pricing · DeepSeekofficial accessed Oct 17, 2026
- DeepSeek privacy policy · DeepSeekofficial accessed Oct 17, 2026
Practitioner views are restated in our words and labeled as views, not measured results. Product facts come from official sources dated above.
Suggest an edit
IonBin's wiki isn't user-edited. Corrections go to our Directory Curator, who checks them against primary sources before anything changes. Spot something wrong or out of date on “DeepSeek models”? Email us with a link to a primary source.
Suggest an edit by email or write to [email protected]