DeepSeek
DeepSeek V4 Pro and V4.1-Flash — 1M context, open-weight, peak/off-peak pricing
DeepSeek's September 2026 API offers two models: DeepSeek V4 Pro (deepseek-v4-pro-0813, frontier) and DeepSeek V4.1-Flash (deepseek-flash, efficient). Peak hours (01:00–10:00 UTC weekdays) are 2× the off-peak rate. Both models support 1M token context, JSON output, tool calls, and the Anthropic API format.
DeepSeek V4 Pro delivers frontier-class intelligence at competitive off-peak rates — open-weight, 1M context, and compatible with both OpenAI and Anthropic SDKs.
- V4 Pro off-peak: competitive per-1M pricing — frontier intelligence at a fraction of Western provider cost
- V4.1-Flash off-peak: fast and cost-efficient — cheapest DeepSeek model tier
- Both models open-weight — can be self-hosted
- Peak/off-peak pricing: off-peak is 50% cheaper than peak
- 1M token context window on both models
- Supports OpenAI and Anthropic API formats
- Data residency in China — may not meet GDPR/compliance requirements
- API reliability can lag Western providers during peak demand
- V4.1-Flash does not support vision inputs
DeepSeek Models
| Model | Modality | Tier | Price | Cached /1M | TTFT | Throughput | Context | Intel. | Caps | |
|---|---|---|---|---|---|---|---|---|---|---|
deepseek-ocr-2 DeepSeek OCR | Chat | budget | $0.03/ $0.03 in / out per 1M EST Jul 2026 | — | 200ms | 300 t/s | 8K | 55 | ||
deepseek-flash DeepSeek V4 | Chat | efficient | $0.15/ $0.60 in / out per 1M EST Oct 2026 | — | 250ms | 220 t/s | 1.0M | 80 | ||
deepseek-r1-distill-llama-70b DeepSeek R1 | Chat | efficient | $0.23/ $0.69 in / out per 1M EST Jul 2026 | — | 900ms | 80 t/s | 66K | 72 | ||
deepseek-chat DeepSeek V3 | Chat | efficient | $0.27/ $1.10 in / out per 1M EST Jul 2026 | $0.07 | 600ms | 90 t/s | 66K | 74 | ||
deepseek-v3-0324 DeepSeek V3 | Chat | efficient | $0.27/ $1.10 in / out per 1M EST Jul 2026 | $0.07 | 550ms | 95 t/s | 66K | 78 | ||
deepseek-v3.1 DeepSeek V3 | Chat | efficient | $0.27/ $1.00 in / out per 1M EST Jul 2026 | — | 450ms | 140 t/s | 164K | 82 | ||
deepseek-v3.1-terminus DeepSeek V3 | Chat | efficient | $0.27/ $1.00 in / out per 1M EST Jul 2026 | — | 460ms | 135 t/s | 164K | 83 | ||
deepseek-v3.2 DeepSeek V3 | Chat | efficient | $0.27/ $0.40 in / out per 1M EST Jul 2026 | — | 430ms | 150 t/s | 164K | 84 | ||
deepseek-v3-turbo DeepSeek V3 | Chat | efficient | $0.40/ $1.30 in / out per 1M EST Jul 2026 | — | 300ms | 200 t/s | 64K | 80 | ||
deepseek-reasoner DeepSeek R1 | Chat | advanced | $0.55/ $2.19 in / out per 1M EST Jul 2026 | $0.14 | 2.2s | 50 t/s | 66K | 82 | ||
deepseek-v4-pro DeepSeek V4 | Chat | advanced | $0.66/ $1.98 in / out per 1M EST Jul 2026 | — | 500ms | 130 t/s | 1.0M | 88 | ||
deepseek-r1-turbo DeepSeek R1 | Chat | advanced | $0.70/ $2.50 in / out per 1M EST Jul 2026 | — | 3.0s | 80 t/s | 64K | 84 | ||
deepseek-prover-v2 DeepSeek Prover | Chat | advanced | $0.70/ $2.50 in / out per 1M EST Jul 2026 | — | 5.0s | 50 t/s | 164K | 86 |
DeepSeek — Frequently Asked Questions (FAQs)
Similar Providers
DeepSeek Reviews
All LLM Providers
Full comparison tableDeepSeek LLM pricing overview
DeepSeek is a proprietary AI lab that develops and hosts its own large language models. DeepSeek's September 2026 API offers two models: DeepSeek V4 Pro (deepseek-v4-pro-0813, frontier) and DeepSeek V4.1-Flash (deepseek-flash, efficient). Peak hours (01:00–10:00 UTC weekdays) are 2× the off-peak rate. Both models support 1M token context, JSON output, tool calls, and the Anthropic API format. Common use cases include Cost-efficiency, Reasoning, Coding, Open-source. Headquartered in Hangzhou, China, founded 2023. All models are accessible via a REST API compatible with standard OpenAI-style request formats, enabling drop-in integration with most LLM frameworks and orchestration tools.
DeepSeek vs other LLM providers
DeepSeek competes with OpenAI, Anthropic, Google, Groq, Together AI, Mistral, Cohere, and other inference API providers across dimensions of price, throughput, latency, context window, and model intelligence. DeepSeek V4 Pro delivers frontier-class intelligence at competitive off-peak rates — open-weight, 1M context, and compatible with both OpenAI and Anthropic SDKs. Use the LLM provider comparison tool to see DeepSeek token pricing, latency, and throughput side-by-side with any other provider. The full LLM pricing table shows all providers ranked by input token cost, output token cost, and throughput in a single sortable view.
Why choose DeepSeek?
DeepSeek stands out because: DeepSeek V4 Pro delivers frontier-class intelligence at competitive off-peak rates — open-weight, 1M context, and compatible with both OpenAI and Anthropic SDKs.. The API currently offers 13 models, spanning 4 advanced, 8 efficient, 1 budget tiers. For teams running high-volume inference workloads, prompt caching and batch API endpoints can reduce effective input token costs by 50–90% — check the context window cost guide for a full breakdown of caching economics.
Understanding DeepSeek token pricing
DeepSeek charges separately for input (prompt) and output (completion) tokens, priced per 1M tokens in USD. Output tokens are typically 3–5× more expensive than input tokens due to the compute cost of autoregressive generation. The cheapest {meta.name} input rate is currently $0.03/1M tokens. Prices shown are sourced from DeepSeek's public pricing page and updated every 15 minutes. Need help estimating your spend? Read the LLM API cost calculator guide — it covers tokens, context windows, prompt caching, and batch discounts with worked examples for RAG, chat history, and document processing workloads.
DeepSeek context window and model capabilities
Context window size directly affects both capability and cost — every token in the context window is charged as an input token. DeepSeek's largest context window is 1.0M tokens. For RAG and document processing workloads, longer context windows enable richer retrieval but increase per-call costs proportionally. Prompt caching — where supported — stores the KV state of repeated prefixes and charges 75–90% less for cache hits, making it the most impactful cost optimization for applications with consistent system prompts or retrieved documents. See the LLM context window cost guide for a full analysis of how context length affects your API bill.
Self-hosted vs managed inference: when DeepSeek makes sense
Managed inference APIs like DeepSeek eliminate infrastructure overhead — no GPU provisioning, driver management, or model serving stack to maintain. The trade-off is cost at scale: a single H100 at ~$2.50/hr can serve ~500K tokens/min of Llama 3.3 70B, which at DeepSeek API rates would cost significantly more per token. The break-even point depends on your request volume, latency requirements, and engineering capacity. For teams processing fewer than ~10M tokens/day, managed APIs are almost always cheaper when total cost of ownership is considered. Above that threshold, self-hosted inference on rented GPU compute typically wins on unit economics. Read the cheapest GPU cloud guide for a full break-even analysis. Historical DeepSeek token price data is available in the LLM price history charts.
Compare DeepSeek With Another LLM Provider
Compare DeepSeek vs other providers
Side-by-side token pricing, latency, throughput, and context window. View all LLM provider comparisons →
Looking for alternatives?
Compare DeepSeek with other LLM API providers side by side.