When you start churning millions of tokens a month, the difference between DeepSeek’s flash models and OpenAI’s GPT‑4o mini becomes stark enough to tip your budget. DeepSeek’s deepseek‑v4‑flash charges $0.0886 for every 1M input tokens, $0.0177 for cached input and $0.1772 for output, while its older deepseek‑chat sits at $0.2574 input and $1.0287 output per 1M tokens. OpenAI’s gpt‑4o‑mini asks $0.15 for input, $0.075 for cached input and $0.6 for output per 1M tokens. In raw numbers the DeepSeek flash tier is roughly half the input price and a third of the output price of GPT‑4o mini, which can translate into sizable savings once you cross the hundred‑million‑token threshold.
### Where the dollars actually go
The table below lays out the exact token rates for the two contenders that matter for high‑throughput workloads. DeepSeek offers a range of flash‑oriented models, but the deepseek‑v4‑flash row is the most comparable to GPT‑4o mini’s latency and feature set. OpenAI’s pricing is a single line for the mini model, with the same rates for the July‑2024 revision.
| Vendor | Model | Input | Cached input | Output |
|---|---|---|---|---|
| DeepSeek | deepseek‑v4‑flash | $0.0886 per 1M tokens | $0.0177 per 1M tokens | $0.1772 per 1M tokens |
| OpenAI | gpt‑4o‑mini | $0.15 per 1M tokens | $0.075 per 1M tokens | $0.6 per 1M tokens |
### Who should care
If your product relies on continuous dialogue or massive batch processing—think customer‑support bots, real‑time analytics, or content generation pipelines—DeepSeek’s flash tier gives you a clear cost edge while still offering cached‑input discounts that matter when you reuse prompts. However, OpenAI’s ecosystem brings built‑in safety layers, broader language coverage, and the brand trust that many enterprises require, which can justify the higher token price for regulated or mission‑critical applications. For startups and scale‑ups that need to keep the burn low and can tolerate a newer provider, DeepSeek’s flash model is the pragmatic pick. For larger organizations where compliance, support contracts, and integration depth outweigh pure token costs, GPT‑4o mini remains the safer bet.
Pricing verified on 2026-09-10.