Chat assistantsDocumented overviewLast checked Oct 5, 2026
DeepSeek offers an AI chat assistant and a usage-priced API for its V4-series language models.
(open full size in a new tab)
(open full size in a new tab)
(open full size in a new tab)
(open full size in a new tab)Our take: Choose DeepSeek when you want to call a 1M-context model through an OpenAI- or Anthropic-compatible API and pay per token, ideally scheduling heavy jobs off-peak.
Main limitation: API prices double during peak hours (01:00-04:00 and 06:00-10:00 UTC on weekdays), and the chat app terms were not verifiable
Choose DeepSeek when you want to call a 1M-context model through an OpenAI- or Anthropic-compatible API and pay per token, ideally scheduling heavy jobs off-peak. Based on the vendor's documentation, not on a hands-on test.
Also a good fit for
Or if
Headlines from US outlets, selected automatically — not endorsements. Updated Oct 5, 2026
Prices checked Oct 5, 2026 on the official US pricing page. Official pricing ↗ Prices and limits change often.
Chat app free status not confirmed; the fetched chat site returned an error to the bot. The API deducts from a granted balance first, then a topped-up balance.
Not documented in fetched pages; granted balance is mentioned in deduction rules.
$0.3 per 1M tokens source (opens in a new tab)
Per 1M tokens
Peak $0.30; off-peak $0.15 (half). Cache hit: $0.006 peak / $0.003 off-peak.
For: Cost-sensitive API use
Limits
Includes
API included
$1.2 per 1M tokens source (opens in a new tab)
Per 1M tokens
Peak $1.20; off-peak $0.60 (half).
For: Cost-sensitive API use
Limits
Includes
API included
$1.32 per 1M tokens source (opens in a new tab)
Per 1M tokens
Peak $1.32; off-peak $0.66 (half). Cache hit: $0.044 peak / $0.022 off-peak.
For: Higher-capability API use
Limits
Includes
API included
$3.96 per 1M tokens source (opens in a new tab)
Per 1M tokens
Peak $3.96; off-peak $1.98 (half).
For: Higher-capability API use
Limits
Includes
API included
See official pricing source (opens in a new tab)
Unknown
For: General chat users
None listed beyond per-token usage; vendor reserves the right to change prices.
Not stated.
DeepSeek is an AI lab whose site offers a web chat with DeepThink and Search modes, plus a developer API platform. The API docs list the deepseek-flash and deepseek-v4-pro models with OpenAI-format and Anthropic-format endpoints, billed per million tokens in USD.
You provide
You get back
Scenarios are worked out by our editors from the documentation; the actual result depends on your input and plan.
Switches the model to its thinking mode before answering; the API docs say deepseek-flash supports both thinking (default) and non-thinking modes. Better answers on multi-step reasoning tasks
Lets the chat assistant look up web results when answering. Fresher answers than the model alone
Exposes the models at https://api.deepseek.com in the OpenAI request format, with JSON output, tool calls and the Responses API. Existing OpenAI-style client code works with a changed base URL
Exposes the same models at https://api.deepseek.com/anthropic in the Anthropic request format. Tools built for the Anthropic API, including agent tools, can point at DeepSeek
You are a support triage assistant. Classify the ticket below as billing, bug or feature_request and return JSON with keys category and one_line_summary. Ticket: "I was charged twice for my October invoice and need a refund."
Export: JSON output (API)
DeepSeek has no public API in its official documentation. OpenAI-format and Anthropic-format base URLs; Agent Integrations section lists coding agents. API documentation ↗
"Better suited" is based on documented features and pricing, not on a head-to-head test.
| Alternative | When to choose it | Pricing | Limitation | |
|---|---|---|---|---|
| ChatGPT | you want a managed chat product over a raw API | See the alternative profile | Different pricing model, not per-token only | Compare |
| Claude | you want Anthropic-hosted models | See the alternative profile | Priced separately | Compare |
| Google Gemini | you already work in Google Workspace | See the alternative profile | Different limits and plans | Compare |
| Grok | you want X-integrated answers | See the alternative profile | Different plans and API | Compare |
No user reviews yet.
Used DeepSeek? Be the first to share your experience.
Off-peak applies to every hour except 01:00-04:00 and 06:00-10:00 UTC on weekdays (Chinese public holidays count as off-peak), and costs half the peak rate. source (opens in a new tab)
Both have 1M context, 384K max output, JSON output and tool calls. Flash supports vision and is cheaper (peak $0.30 input cache-miss, $1.20 output); Pro does not support vision (peak $1.32 / $3.96). source (opens in a new tab)
Yes. The docs list https://api.deepseek.com for the OpenAI format and https://api.deepseek.com/anthropic for the Anthropic format. source (opens in a new tab)
Cost equals tokens times price, deducted from a granted balance first and then your topped-up balance. source (opens in a new tab)
They are still accepted, but the old models are retired; requests are served by DeepSeek-V4.1-Flash and billed at the Flash price. source (opens in a new tab)
Not verified: Chat app free status and plans; Privacy policy: training use, retention, commercial use; Mobile/Desktop availability; Account requirement details; US availability.
Every fact about DeepSeek above comes from the 3 official pages listed here, last compared on Oct 5, 2026; we have not run DeepSeek ourselves, so it has no editorial score. How we check and re-check pages.
Article updated Oct 5, 2026; prices checked Oct 5, 2026. Outbound links to vendors are plain links; we do not earn commissions. How we review · Report an error
compare>