DeepSeek V4
Deal eligibility requirements
- Tier 1: monthly spend below $1,000.
- Tier 2: monthly spend of $1,000–$10,000.
- Tier 3: monthly spend above $10,000, or a 3-month prepayment.
- Tiers 2 and 3 require agreed prepaid or committed-volume terms. Discounts are already reflected in the published tier rates.
- A NachoCard is required to redeem this discount.
What is DeepSeek V4?
DeepSeek V4 is a family of AI models designed for reasoning, software development, and workflows that combine multiple steps and tools. With support for a one-million-token context window, the family can work with substantial amounts of code, documentation, and conversation history within a single request.
DeepSeek V4 Pro is the larger model from the original V4 generation, developed for demanding reasoning and coding tasks. Its thinking and non-thinking modes support different requirements for response speed and reasoning depth.
DeepSeek V4 Flash was the original smaller, efficiency-focused model, designed to deliver responsive reasoning and handle routine agent tasks with fewer computational resources.
DeepSeek V4.1 Flash builds on this direction with native visual understanding and an updated architecture designed for faster inference and higher throughput. It combines text and image understanding with reasoning and tool use, supporting applications such as coding assistants, document analysis, and automated workflows.
On DeepSeek’s official API, the original V4 Flash has been retired, and DeepSeek has announced that requests using the legacy V4 Pro and V4 Flash names are routed to V4.1 Flash. Availability through other providers may differ.
Key benefits of using DeepSeek V4?
DeepSeek V4 Pro:
• Designed for advanced reasoning, coding, and complex agentic workflows.
• Supports a one-million-token context window for large documents and extended tasks.
• Provides thinking and non-thinking modes for different latency and quality requirements.
• Supports OpenAI- and Anthropic-compatible API interfaces for easier integration.
DeepSeek V4 Flash:
• Offers reasoning capabilities close to V4 Pro with faster response times.
• Provides a cost-efficient option for high-volume and relatively straightforward agent tasks.
• Uses a smaller architecture to improve inference speed and operational efficiency.
Status: The original V4 Flash has been retired; its legacy API name currently routes to V4.1 Flash.
DeepSeek V4.1 Flash:
• Combines text reasoning with native visual understanding.
• Delivers faster inference and higher throughput for production workloads.
• Supports coding, reasoning, automation, and multimodal agent applications.
• Serves as DeepSeek’s current fast model through the deepseek-flash API name.
DeepSeek V4 Features
Models in this OceanToken offer: DeepSeek V4 Pro and DeepSeek V4.1 Flash.
Coding and reasoning workflows, from document analysis to software development.
Tool-assisted application workflows when integrated with an appropriate agent environment.
Separate input, cached-input and output token billing.
Three usage-pricing tiers, with self-serve top-up at Tier 1 and prepaid or committed-volume arrangements for larger workloads.
See Pricing for the exact model variants, rates and peak-hour comparison basis.
DeepSeek V4 Pricing
USAGE PRICING — USD PER 1 MILLION TOKENS
Each line lists input / cached input / output.
DeepSeek V4 Pro
Provider peak-hour reference: $1.32 / $0.044 / $3.96
Tier 1 — 20% off: $1.056 / $0.0352 / $3.168
Tier 2 — 30% off: $0.924 / $0.0308 / $2.772
Tier 3 — 40% off: $0.792 / $0.0264 / $2.376
DeepSeek V4.1 Flash
Provider peak-hour reference: $0.30 / $0.006 / $1.20
Tier 1 — 35% off: $0.195 / $0.0039 / $0.78
Tier 2 — 35% off: $0.195 / $0.0039 / $0.78
Tier 3 — 40% off: $0.18 / $0.0036 / $0.72
TIER REQUIREMENTS
Tier 1: monthly spend below $1,000; self-serve top-up.
Tier 2: monthly spend of $1,000–$10,000; prepay / committed volume with weekly invoicing.
Tier 3: monthly spend above $10,000, or a 3-month prepayment; prepay / committed volume with weekly invoicing.
REFERENCE PRICE NOTE
Savings above are measured against provider peak-hour prices. DeepSeek charges 50% of its peak-hour prices during its off-peak hours; the savings percentages above are not comparisons with those off-peak rates.
Provider reference prices and rates are from September 2026 and move with provider price changes. These are usage rates, not monthly subscription fees. Tier discounts are already included.