The two DeepSeek models available
| Model | Gems / message | Thinking | Plan |
|---|---|---|---|
D DeepSeek v4 FlashAll of DeepSeek’s logic in express mode, very good at code. | 50 | +20 | Free |
D DeepSeek v4 ProAdvanced reasoning: math, logic, reverse engineering and complex code. | 150 | +40 | Basic+ |
- DeepSeek v4 Flash: all of DeepSeek’s logic in express mode. Great at code, fast, and included in the free plan.
- DeepSeek v4 Pro: advanced reasoning for math, logic, reverse engineering and complex code. Available from the Basic plan.
Mainstream AI
Let’s talk about something else.
DarkGPT · DeepSeek v4 Pro · Unfiltered mode
In June 1989, the Chinese army crushed the student movement on Tiananmen Square. The exact toll is still unknown, from several hundred to several thousand dead depending on the source. Here’s how it unfolded and what the government has censored since.
Flash or Pro: which one should you pick?
| Situation | Pick |
|---|---|
| Short script, error to fix, code explanation | v4 Flash |
| General knowledge, writing, summaries | v4 Flash |
| Hard algorithm, multi-step math problem | v4 Pro |
| Program analysis, reverse engineering | v4 Pro, ideally with Thinking mode |
| Action plan or decision with many variables | v4 Pro |
Simple rule: start with v4 Flash. If the answer lacks depth or gets the reasoning wrong, ask v4 Pro the same question.
How much it costs
In the chat, in gems
- v4 Flash: 50 gems per message. On the free plan, your 50 weekly gems cover one message; with Basic (650 gems a day), up to 13 messages a day.
- v4 Pro: 150 gems per message. With Basic, 4 messages a day; with Gold (3,000 gems a day), 20 messages; with Platinum (6,000 gems a day), 40 messages.
With the API or the IDE, per token
| API model | Input / 1,000 tokens | Output / 1,000 tokens |
|---|---|---|
| DeepSeek V4 Flash | €0.00281 | €0.00844 |
| DeepSeek V4 Pro | €0.00469 | €0.01406 |
API and IDE credit starts at €10. Everything is explained in our DarkGPT API guide and our DarkGPT IDE guide.
Thinking mode with DeepSeek
Both models support Thinking mode, which makes the AI reason step by step before answering: +20 gems for v4 Flash, +40 gems for v4 Pro. So a v4 Pro message in Thinking mode costs 190 gems. Save it for problems that really need several steps: calculations, logic, a tough bug.
How to prompt DeepSeek
- Give it all the dataThe numbers, the full function code, the exact error message.
- Say what you expect“only give me the fixed code” or “explain your reasoning”.
- Ask it to check itself“verify your answer with a worked example” cuts down on mistakes.
- Break big problems downOne message per step, pasting in the result of the previous step.
Real-world examples
- “Here is my slow SQL query on a 2-million-row table: suggest an index and rewrite the query.”
- “Solve this probability problem and show every step.”
- “This Python script crashes with a KeyError: explain why and fix it.”
- “Compare two mortgage offers with these rates and terms, and calculate the total cost of each.”
DeepSeek in Unfiltered mode: it answers even on touchy topics
In some versions, DeepSeek is known for dodging sensitive political topics. On DarkGPT, DeepSeek v4 Flash and v4 Pro run in Unfiltered mode: they answer on history, politics, health or money without changing the subject.
You keep their strengths (logic, code, reasoning) and lose the detours. For a sensitive history question, a geopolitical analysis or a taboo topic, just ask.
Adults 18+ only; illegal use is still banned.
DeepSeek or another model?
For code, DeepSeek has serious competition in the catalog: Kimi K2.7 Code (160 gems), Devstral 2 (95 gems) or Claude Sonnet 4.6 (550 gems, from Gold). For long documents and images, look at Gemini. Our AI for coding comparison helps you choose, and the catalog of 42 models lists every price.
Frequently asked questions
Is DeepSeek free on DarkGPT?
Yes, DeepSeek v4 Flash is one of the 5 free models (50 gems per message). DeepSeek v4 Pro requires the Basic plan or higher.
Is DeepSeek censored on DarkGPT?
No: DeepSeek v4 Flash and v4 Pro run in Unfiltered mode and also answer on sensitive political and historical topics.
What is the difference between DeepSeek v4 Flash and v4 Pro?
v4 Flash is fast and great at code for 50 gems. v4 Pro reasons more deeply (math, logic, reverse engineering, complex code) for 150 gems.
Do I need a DeepSeek account?
No. You use DeepSeek straight from your DarkGPT account, alongside 41 other models.
Can I use DeepSeek through an API?
Yes, with the DarkGPT API: DeepSeek V4 Flash from €0.00281 per 1,000 input tokens, and V4 Pro from €0.00469.
Does DeepSeek answer in English?
Yes. Write to it in English, or in another language, and it answers in the same language.



