DeepSeek API Platform
API access to DeepSeek's chat and reasoning models.
Find alternatives to DeepSeek API Platform
What it does
- Model inference
- Text generation
- Code generation
Features 11
- JSON OutputOption to make the model output valid JSON strings for structured output.
- Thinking ModeModel outputs chain-of-thought reasoning before the final answer to improve accuracy.
- Context CachingDisk-based prefix caching enabled by default for all users with no code changes.
- FIM Completion (Beta)Fill-in-the-middle completion from a prefix and optional suffix, used for code completion.
- Files APIUploads images and references them later by file_id across requests.
- Anthropic API formatSupports the Anthropic API format via a dedicated base URL.
- VisionThe deepseek-flash model accepts images alongside text for description and analysis.
- Tool CallsLets the model call external tools to extend its capabilities when using the DeepSeek API.
- Responses API formatThe API supports the Responses API format, added to meet demand for use with Codex.
- Search & Fetch APISearch API and Fetch API give agents and RAG applications access to live web content.
- Chat Prefix CompletionUsers provide an assistant prefix message and the model completes the rest of the message.
How it compares
-
K-EXAONE 2.0
It also delivered performance comparable to leading global models, including GLM-5.1, DeepSeek V4 Pro Max and Qwen3.5, in instruction-following benchmarks.
News reporting · 19 Sep 2026 -
Solar LLMs
Solar Pro 2 delivered results comparable to leading frontier-scale models including OpenAI's GPT-4o, DeepSeek's R1, Mistral's Small 3.2, and Alibaba's Qwen 3—despite being less than half their size.
Official documentation · 19 Sep 2026
Sources
- Pricing
-
Models & Pricing page: "We bill model usage based on the total number of input and output tokens by the model." Rates are listed per 1M tokens for deepseek-flash and deepseek-v4-pro, and "Search and Fetch are standalone APIs billed per request."
- Security
-
Official documentation · 18 Oct 2026
DeepSeek privacy policy lists, among its purposes, training its models by monitoring interactions and usage. It states no certification, retention mode or residency, so those stay empty or unknown.
- Sold within
-
DeepSeek's Models & Pricing page publishes per-model token rates: "The prices listed below are in units of per 1M tokens." Fees "will be directly deducted from your topped-up balance or granted balance".
- Interfaces
-
Official documentation · 28 Sep 2026
Docs give the api.deepseek.com endpoint and a curl example with a bearer API key.
- Modalities
-
Official documentation · 1 Oct 2026
Page: 'an array of content parts (for image input) / Creates a model response for the given chat conversation'.
- Intended for
-
Official documentation · 13 Oct 2026
Page: 'The DeepSeek API uses an API format compatible with OpenAI/Anthropic'.
Also from DeepSeek
-
DeepSeek Application
Assistant app for chat, reasoning and code questions, built on DeepSeek's own models.
-
DeepSeek Harness Developer tool
Open-source agent harness built from composable plugins, now with a macOS and Windows desktop app plus npx web install and GitHub source; preview release.
Open DeepSeek in the directory
Something wrong here? Send a correction — quote this product id: deepseek-api-platform.