TechCentral.ie · Pro · Republished as an independent data story
DeepSeek announces steep price increase due to rising demand for low-cost AI models
Announcement comes as DeepSeek-V4-Flash-0731 becomes fastest-growing model ever by token usage
DeepSeek has announced that it will significantly increase the prices of its API services in the near future.
According to the Chinese AI start-up, the new rates will be disclosed later, but users are advised to already take this into account in their usage. The announcement comes a week after the introduction of DeepSeek-V4-Flash-0731, a lightweight AI model with 284 billion parameters.
The model attracted worldwide attention because it delivers performance that comes close to the best AI models, while costing much less. The price increase highlights the challenge DeepSeek faces: a company that built its reputation on very low prices now has to maintain that strategy in a market where competition is increasingly fierce.
01 · The announcement
Why DeepSeek is raising prices
DeepSeek built its reputation on rock-bottom AI pricing. Its lightweight models deliver performance that comes close to the best in the world — for a fraction of the cost. Now the company says demand has risen so much that API prices will increase significantly in the near future, with new rates to be disclosed later.
Users are advised to take this into account in their usage now. The move is a turning point for a company whose rock-bottom pricing changed the economics of AI — and the announcement lands just one week after the launch of DeepSeek-V4-Flash-0731, the model that made the demand impossible to ignore.
02 · Cost comparison
About 3 cents vs $3.15: the 105× gap
According to Artificial Analysis, an average task with DeepSeek-V4-Flash-0731 costs about 3c (US), compared to $3.15 US dollars for Anthropic's Claude Fable 5. That makes the DeepSeek model roughly 105 times cheaper.
03 · Per-million-token pricing
DeepSeek vs Zhipu AI, per million tokens
The current API prices are 14c per million input tokens and 0.28c per million output tokens, while competitor Zhipu AI charges $1.40 and $4.40 respectively.
| Provider | Input (per 1M tokens) | Output (per 1M tokens) |
|---|---|---|
| DeepSeek-V4-Flash-0731 | 0¢ | 0.28¢ |
| Zhipu AI | $1.40 | $4.40 |
A million output tokens on Zhipu AI costs about 1,571 times the DeepSeek output price — DeepSeek is not only cheaper than American competitors, but also much more affordable than comparable Chinese models.
04 · Open-weight ranking
The most powerful open-weight AI models
Research firm Epoch AI currently calls DeepSeek-V4-Flash-0731 the second most powerful open-weight AI model on the market, only surpassed by the Kimi K3 model from China's Moonshot AI.
05 · Performance positioning
Between Claude Opus 4.5 and Opus 4.6
Its performance is said to be somewhere between Anthropic's Claude Opus 4.5 and Opus 4.6.
06 · Interactive calculator
What would your monthly bill look like?
Adjust monthly token volume and see DeepSeek's current rates against competitors, updating in real time.
Drag to set input tokens per month (in millions). Output is assumed at 1/10 of input.
07 · How it works
The engineering behind low prices
According to Professor Zhenhui Jack Jiang of the University of Hong Kong, who spoke to the South China Morning Post, DeepSeek owes its low prices to an efficient technical approach.
The mixture-of-experts idea: only part of the network is activated for each task. Less computing power and memory are required.
Mixture-of-experts architecture. Only part of the network is activated for each task, so less computing power and memory are required.
Efficient training. DeepSeek uses a training method in which knowledge from multiple specialised AI models is combined into a single model, allowing it to be trained more efficiently.
The result. A lightweight 284-billion-parameter model that approaches frontier performance — at a fraction of the cost.
08 · Voices
The analysts and the experts
"A price increase at this moment could be risky, because new models from, among others, Meta and OpenAI deliver comparable performance at competitive rates."
"DeepSeek owes its low prices to an efficient technical approach — a mixture-of-experts architecture in which only part of the network is activated for each task, requiring less computing power and memory."
Growing fast where models run locally
On the Ollama platform, which allows AI models to be run locally, DeepSeek-V4-Flash-0731 has, according to the company, become the fastest-growing model ever measured in terms of token usage. Ollama is therefore expanding capacity in the United States and Europe.
#1 fastest-growing model by token usage US + EU capacity expansion
09 · Timeline
How the story unfolded
-
10 August 2026 · −1 week
DeepSeek-V4-Flash-0731 launches
Lightweight model with 284 billion parameters arrives — performance close to the best models, at far lower cost.
-
August 2026
Fastest-growing model ever on Ollama
Token usage on the local-run platform breaks records; Ollama expands capacity in the United States and Europe.
-
10 August 2026
Price announcement
DeepSeek says API prices will increase significantly in the near future; new rates to be disclosed later.
-
What's next
Rates disclosed — and a strategy test
New rates will be disclosed later. Can DeepSeek keep its rock-bottom reputation in a fiercely competitive market?
10 · What this means
A turning point for AI's price war
For developers
API prices will rise significantly in the near future. If DeepSeek's rates are part of your cost base, the new rates — to be disclosed later — deserve attention now.
For the market
A company famous for rock-bottom pricing is raising prices. Rivals like Meta and OpenAI are ready with comparable performance at competitive rates, and AI developers are watching closely.
For open-weight AI
DeepSeek-V4-Flash-0731 remains the #2 most powerful open-weight model behind Kimi K3 — the economics that made it the fastest-growing model ever measured on Ollama now face their first real test.
11 · Glossary
Terms, explained
- Tokens
- The units a language model reads and writes. One input token is roughly a word or part of a word; prices are quoted per million tokens.
- Open-weight models
- Models whose trained weights are publicly released, so anyone can run, fine-tune, and study them — typically with far lower costs than proprietary rivals.
- API
- Application Programming Interface — the way developers call a model over the internet and pay per token used.
- Mixture-of-experts
- An architecture in which only part of the network is activated for each task, requiring less computing power and memory.
12 · FAQ
Questions, answered
How much is DeepSeek raising prices?
DeepSeek has announced the prices of its API services will increase significantly in the near future. The new rates have not been disclosed yet — they will be revealed later.
Why is the company raising prices now?
The official driver is rising demand for its low-cost models — DeepSeek-V4-Flash-0731 has become the fastest-growing model ever measured by token usage on Ollama. The announcement lands one week after that model's launch.
How cheap is DeepSeek-V4-Flash-0731 really?
An average task costs about 3c (US) — compared with $3.15 US dollars for Anthropic's Claude Fable 5, making the DeepSeek model roughly 105 times cheaper, according to Artificial Analysis.
How powerful is the model?
Epoch AI calls it the second most powerful open-weight AI model on the market, surpassed only by Kimi K3 from Moonshot AI. Performance is said to be between Anthropic's Claude Opus 4.5 and Opus 4.6.
How does DeepSeek keep costs so low?
Professor Zhenhui Jack Jiang of the University of Hong Kong points to an efficient technical approach: a mixture-of-experts architecture that activates only part of the network per task, plus a training method that combines knowledge from specialised models into a single one.
Is this DeepSeek's official website?
No. This page is an independent presentation based on TechCentral.ie reporting by Niall Kitson, published 10 August 2026. It is not affiliated with DeepSeek.
13 · Sources
Verify everything yourself
- TechCentral.ie — "DeepSeek announces steep price increase due to rising demand for low-cost AI models" by Niall Kitson (original article)
- Epoch AI — open-weight model rankings
- Artificial Analysis — cost-per-task comparisons
- South China Morning Post — interview with Professor Zhenhui Jack Jiang
- Ollama — local model platform and token-usage growth
- X — Michael Guo's warning post
Independent presentation. Based on TechCentral.ie reporting by Niall Kitson, 10 August 2026, published at TechCentral.ie. This is not DeepSeek's official site.