deepseek-v4-pro vs kimi-k3.1-pro: pricing and capabilities compared
In each model's cheapest group on this site, deepseek-v4-pro costs ¥16.8 input and ¥33.6 output (China Models group) and kimi-k3.1-pro costs ¥18.2 input and ¥95.19 output (China Models group), per 1M tokens.
deepseek-v4-pro is cheaper on input, output and cache reads, by 8%, 65% and 69% respectively.
deepseek-v4-pro is offered in 2 groups and kimi-k3.1-pro in 2 groups. Groups offering both: Standard and China Models; a key in any of them can call either model.
Prices updated Nov 13, 2026, 6:08 PM
At a glance
| Item | deepseek-v4-pro | kimi-k3.1-pro |
|---|---|---|
| Vendor | DeepSeek | Moonshot Kimi |
| Billing | Per token | Per token |
| Cheapest group | China Models | China Models |
| Input ¥ per 1M tokens | ¥16.88% lower | ¥18.2 |
| Output ¥ per 1M tokens | ¥33.665% lower | ¥95.19 |
| Cache read ¥ per 1M tokens | ¥0.56169% lower | ¥1.82 |
| Groups | 2 groups | 2 groups |
| Input types | — | text and images |
| Output types | — | text |
| Capability tags | Vision | Vision, Multimodal, Tool use, Reasoning and File analysis |
| Endpoints | Chat Completions | Chat Completions |
| Available since | Aug 22, 2026 | Oct 24, 2026 |
Prices are each model's cheapest group, in CNY; the side marked "lower" is cheaper. Data comes from our price list and model catalog; anything not listed shows as —.
Prices in every group
The same model costs different amounts in different groups; a key's group decides its line and its price.
| Group | Input ¥ per 1M tokens | Output ¥ per 1M tokens | Cache read ¥ per 1M tokens |
|---|---|---|---|
| StandardBoth | ¥24 | ¥48 | ¥0.802 |
| China ModelsCheapestBoth | ¥16.8 | ¥33.6 | ¥0.561 |
| Group | Input ¥ per 1M tokens | Output ¥ per 1M tokens | Cache read ¥ per 1M tokens |
|---|---|---|---|
| StandardBoth | ¥26 | ¥135.98 | ¥2.6 |
| China ModelsCheapestBoth | ¥18.2 | ¥95.19 | ¥1.82 |
How to choose
These points use each model's cheapest-group prices and catalog data. Before going live, run real requests with a key in your target group and check quality, speed and billing.
- On unit prices alone (each at its cheapest group): deepseek-v4-pro is cheaper on input, output and cache reads, so the same token usage costs less.
- When you resend the same long prefix (a fixed system prompt, the same long document), cached input is billed at the cache-read price: deepseek-v4-pro ¥0.561, kimi-k3.1-pro ¥1.82 (per 1M tokens).
- Both support Chat Completions, so requests look the same and switching means changing the model parameter (the key's groups must offer both models).
How to call them
Both models work with a Tutujin API key: set your SDK's base_url to https://www.tutujin.com/v1 and use the model name; everything else is written as with the official SDK.
- deepseek-v4-pro: prices in every group, availability and API docs
- kimi-k3.1-pro: prices in every group, availability and API docs
- Every DeepSeek model and its price
- Every Moonshot Kimi model and its price
- Chat Completions API reference
- Quickstart: create a key and send your first request
- How to choose a group
- Models page: every model with prices in every group
- How billing works: pay as you go, groups and balance
More comparisons
- gpt-5.5 vs deepseek-v4-pro
- gpt-5.5 vs kimi-k3.1-pro
- claude-opus-5-5 vs deepseek-v4-pro
- claude-opus-5-5 vs kimi-k3.1-pro
- gpt-6-sol vs deepseek-v4-pro
- gpt-6-sol vs kimi-k3.1-pro
- gemini-4-carbon vs deepseek-v4-pro
- gemini-4-carbon vs kimi-k3.1-pro
- deepseek-v4-pro vs qwen3.8-max
- deepseek-v4-pro vs glm-5.3
- qwen3.8-max vs kimi-k3.1-pro
- kimi-k3.1-pro vs glm-5.3
FAQ
Which is cheaper, deepseek-v4-pro or kimi-k3.1-pro?
In each model's cheapest group on this site, deepseek-v4-pro costs ¥16.8 input and ¥33.6 output (China Models group) and kimi-k3.1-pro costs ¥18.2 input and ¥95.19 output (China Models group), per 1M tokens. deepseek-v4-pro is cheaper on input, output and cache reads, by 8%, 65% and 69% respectively. What you actually pay depends on the split between input and output tokens and on the group you use. Prices updated Nov 13, 2026, 6:08 PM.
How do I call deepseek-v4-pro and kimi-k3.1-pro?
Create an API key in the console, set your SDK's base_url to https://www.tutujin.com/v1 and use deepseek-v4-pro or kimi-k3.1-pro as the model. deepseek-v4-pro supports: Chat Completions. kimi-k3.1-pro supports: Chat Completions.
Which groups offer deepseek-v4-pro and kimi-k3.1-pro?
deepseek-v4-pro: Standard and China Models. kimi-k3.1-pro: Standard and China Models. Groups offering both: Standard and China Models; a key in any of them can call either model.
Where do the prices on this page come from?
From our live price list — the same data as the model pages and the models page — shown for each model's cheapest group. Prices updated Nov 13, 2026, 6:08 PM.