DeepSeek V4 Pro 0813
- Input / 1M tokens
- $0.11
- Output / 1M tokens
- $4.44
- Providers
- 20
- Uptime, 24 h
- 100%
API models · Side by side
DeepSeek V4 Pro vs DeepSeek V4.1 Flash
DeepSeek V4 Pro and DeepSeek V4.1 Flash come from the same company. Both are open-weight under the MIT licence, read about a million tokens, and are sold by many providers. Is Pro worth paying more for? Here is how they differ on price, inputs, limits and self-hosting.
DeepSeek V4.1 Flash
- Input / 1M tokens
- $0.30
- Output / 1M tokens
- $1.20
- Providers
- 28
- Uptime, 24 h
- 100%
The short answer
At every provider that sells both models, V4.1 Flash costs less than V4 Pro for input and for output. Flash also reads images, while Pro reads text only. Pro is the far larger model, at about 1.65 trillion parameters against about 763 billion for Flash. We compare the August 2026 release of V4 Pro, which replaces the April preview.
Estimate your monthly bill
Enter your own usage. We apply each model's current rates, including long-prompt pricing, for a 30-day month.
Try:
DeepSeek V4 Pro 0813$73.26per month
DeepSeek V4.1 Flash$36.00per month
A token is roughly three-quarters of an English word. Cached input, reasoning tokens, tools and media are not included.
Compare the details
Compare context, supported inputs and usage costs. Highlighted rows show a difference.
| Feature | DeepSeek V4 Pro 0813 | DeepSeek V4.1 Flash |
|---|---|---|
| Input formats | text | image, text |
| Output formats | text | text |
| Context / position limit | 1,048,576 | 1,048,576 |
| Endpoint output limit (tokens) | 943,718 | 943,718 |
| Developer features | frequency penalty, Reasoning output, logit bias, logprobs, Output limit, min p, presence penalty, Reasoning controls, Thinking effort, repetition penalty, Response format, Seed, Stop sequences, Structured output, Temperature, Tool selection, Tool calling, top k, top logprobs, Sampling controls | frequency penalty, Reasoning output, logit bias, logprobs, Output limit, min p, parallel tool calls, presence penalty, Reasoning controls, Thinking effort, repetition penalty, Response format, Seed, Stop sequences, Structured output, Temperature, Tool selection, Tool calling, top a, top k, top logprobs, Sampling controls |
| Cached input / 1M tokens | $0.10 | $0.006000 |
| USD input / 1M tokens | $0.11 | $0.30 |
| USD output / 1M tokens | $4.44 | $1.20 |
| Available API providers | 20 | 28 |
| Best provider uptime, last 24 h | 100% | 100% |
| Price range across providers (per 1M tokens) | Input $0.100–$1.65, output $1.98–$8.00 across 20 providers | Input $0.040–$3.00, output $0.42–$12.00 across 28 providers |
| Model identifier | deepseek/deepseek-v4-pro-0813 | deepseek/deepseek-v4.1-flash |
API pricing
Prices are in USD per million tokens. Input is what you send; output is what the model generates. Chat subscriptions and API usage are billed separately.
Tools, images and additional reasoning can add charges. The examples below cover input and output tokens only.
Compare local modelsWhat those prices mean in practice
Each example uses the same token budget for both models. These are calculations, not measured workloads. Cache discounts, media, tools and extra reasoning are excluded.
| Workload | DeepSeek V4 Pro 0813 | DeepSeek V4.1 Flash |
|---|---|---|
| 1,000 short requestsPer request: 2,000 input + 500 output tokens | $2.44 | $1.20 |
| 100 document summariesPer request: 20,000 input + 1,000 output tokens | $0.67 | $0.72 |
| 10 long-document requestsPer request: 300,000 input + 2,000 output tokens | $0.42 | $0.92 |
Estimate = requests × (input tokens × input price + output tokens × output price) ÷ 1,000,000. Long-prompt rates are applied where the pricing data provides them. Real requests can use different amounts of output.
Beyond the price tag
Price depends on the provider
Neither model has one fixed price. About 21 providers sell V4 Pro and 32 sell V4.1 Flash, so the OpenRouter headline price is just the cheapest host that day. At each of the 17 providers that offer both, Flash is cheaper than Pro for input and output. One quirk: DeepSeek's own V4 Pro endpoint charges double in two weekday windows, 01:00 to 04:00 and 06:00 to 10:00 UTC. Use the live price range and calculator on this page with your own token counts.
Inputs, output limits and controls
V4.1 Flash accepts text and images. V4 Pro accepts text only. Both return text, support tool calling and structured output, and let you set a reasoning effort level. Output caps depend on the provider: about 16K to 944K tokens for Pro and about 131K to 944K for Flash.
Running them yourself
Both are free to download under the MIT licence, with no revenue or usage conditions. V4 Pro is about 1.65 trillion parameters, with about 49 billion active per token, so plan for a large multi-GPU cluster. V4.1 Flash is about 763 billion parameters, with 8 to 16 billion active per token. That is under half Pro's size, but still a multi-GPU job.
Which one should you pick?
Choose DeepSeek V4.1 Flash if you need image input, want the lower bill at any shared provider, or want the smaller model to self-host. Choose DeepSeek V4 Pro if you work with text only and want to test the much larger model on your hardest tasks. Either way, run both on your own prompts before committing.
Common questions
Is DeepSeek V4 Pro always more expensive than V4.1 Flash?
At every provider that sells both, yes, for both input and output. Across different providers the ranges overlap, so a cheap Pro host can undercut an expensive Flash host. Check the live range on this page.
Which DeepSeek V4 Pro does this page compare?
The August 2026 release, listed on OpenRouter as DeepSeek V4 Pro 0813. DeepSeek describes it as the official release that supersedes the April preview.
Can I run either model on one GPU?
No. Both need a multi-GPU cluster, though Flash is under half Pro's size. Our model hardware calculator can size a setup for you.