DeepSeek
- Lowest price / 1M tokens
- $0.090 in · $0.18 out
- Lowest-priced model
- DeepSeek V4 Flash 0423
- Highest-priced model
- DeepSeek V4 Pro 0423 ($1.91 out)
- Largest context
- 1,048,576 tokens
- Also accepts
- image
- Downloadable weights
- 6 of 6 models
AI brands · Side by side
DeepSeek vs ChatGPT
DeepSeek is made by DeepSeek and ChatGPT is made by OpenAI. We compare the models behind them through their APIs: the current lineup, price per million tokens, context limits and the inputs each one accepts. Prices checked 13 November 2026.
ChatGPT
- Lowest price / 1M tokens
- $0.10 in · $0.50 out
- Lowest-priced model
- GPT-6 Luna
- Highest-priced model
- GPT-6.1 Sol ($10.00 out)
- Largest context
- 1,050,000 tokens
- Also accepts
- file, image
- Downloadable weights
- 0 of 6 models
The short answer
Through the API, DeepSeek’s lowest-priced current model, DeepSeek V4 Flash 0423, costs $0.090 per million input tokens and $0.18 per million output tokens; ChatGPT’s lowest-priced current model, GPT-6 Luna, costs $0.10 per million input tokens and $0.50 per million output tokens. DeepSeek is the cheaper place to start. At the top end, DeepSeek V4 Pro 0423 costs $1.91 per million output tokens and GPT-6.1 Sol costs $10.00 per million output tokens.
DeepSeek models accept up to 1,048,576 tokens, ChatGPT models accept up to 1,050,000 tokens. The largest limits are about the same. The limit can differ between models in a family, so check the one you plan to use.
Current models, side by side
The newest text models each company lists through the API, with batch and dated snapshot versions left out. Prices are in USD per million tokens.
DeepSeek models
| Model | Input / 1M | Output / 1M | Context (tokens) | Also accepts | Weights |
|---|---|---|---|---|---|
| DeepSeek V4.1 Flash | $0.30 | $1.20 | 1,048,576 | image | Downloadable |
| DeepSeek V4 Pro 0423 | $0.96 | $1.91 | 1,048,576 | Text only | Downloadable |
| DeepSeek V4 Flash 0423 | $0.090 | $0.18 | 1,048,576 | Text only | Downloadable |
| DeepSeek V3.2 | $0.26 | $0.38 | 163,840 | Text only | Downloadable |
| DeepSeek V3.1 Terminus | $0.27 | $1.00 | 163,840 | Text only | Downloadable |
| DeepSeek V3.1 | $0.25 | $0.95 | 163,840 | Text only | Downloadable |
ChatGPT models
| Model | Input / 1M | Output / 1M | Context (tokens) | Also accepts | Weights |
|---|---|---|---|---|---|
| GPT-6.1 Terra Pro | $1.00 | $5.00 | 1,050,000 | file, image | API only |
| GPT-6.1 Terra | $1.00 | $5.00 | 1,050,000 | file, image | API only |
| GPT-6.1 Sol Pro | $2.00 | $10.00 | 1,050,000 | file, image | API only |
| GPT-6.1 Sol | $2.00 | $10.00 | 1,050,000 | file, image | API only |
| GPT-6 Luna Pro | $0.10 | $0.50 | 1,050,000 | file, image | API only |
| GPT-6 Luna | $0.10 | $0.50 | 1,050,000 | file, image | API only |
What those prices mean in practice
The same token budget for each model. These are calculations, not measured workloads; cache discounts, media, tools and extra reasoning are left out.
Lowest-priced model from each
| Workload | DeepSeek V4 Flash 0423 | GPT-6 Luna |
|---|---|---|
| 1,000 short requestsPer request: 2,000 input + 500 output tokens | $0.27 | $0.45 |
| 100 document summariesPer request: 20,000 input + 1,000 output tokens | $0.20 | $0.25 |
| 10 long-document requestsPer request: 300,000 input + 2,000 output tokens | $0.27 | $0.61 |
Highest-priced model from each
| Workload | DeepSeek V4 Pro 0423 | GPT-6.1 Sol |
|---|---|---|
| 1,000 short requestsPer request: 2,000 input + 500 output tokens | $2.87 | $9.00 |
| 100 document summariesPer request: 20,000 input + 1,000 output tokens | $2.10 | $5.00 |
| 10 long-document requestsPer request: 300,000 input + 2,000 output tokens | $2.90 | $12.30 |
Estimate = requests × (input tokens × input price + output tokens × output price) ÷ 1,000,000, using long-prompt rates where the provider lists them.
Common questions
Is DeepSeek or ChatGPT cheaper?
Through the API, DeepSeek’s lowest-priced current model, DeepSeek V4 Flash 0423, costs $0.090 per million input tokens and $0.18 per million output tokens; ChatGPT’s lowest-priced current model, GPT-6 Luna, costs $0.10 per million input tokens and $0.50 per million output tokens. DeepSeek is the cheaper place to start. At the top end, DeepSeek V4 Pro 0423 costs $1.91 per million output tokens and GPT-6.1 Sol costs $10.00 per million output tokens.
Which has the bigger context window?
DeepSeek models accept up to 1,048,576 tokens, ChatGPT models accept up to 1,050,000 tokens. The largest limits are about the same. The limit can differ between models in a family, so check the one you plan to use.
Which is better, DeepSeek or ChatGPT?
It depends on the task. This page compares price, limits and supported inputs from each provider’s API listing; it does not score answer quality. Run the same prompts through the models you are considering. The cost tables show what a test run costs.
Is the DeepSeek or ChatGPT subscription the same as the API?
No. The chat apps and their subscriptions are billed separately from the API. The prices on this page are API prices per million tokens, which is what you pay when you build the models into your own application.
Can I run DeepSeek on my own server?
6 of the 6 current DeepSeek models listed here have downloadable weights. ChatGPT is API-only. Large models need several high-memory GPUs, so check the hardware calculator before planning a server.
Planning to run it yourself?
Downloadable weights still need enough GPU memory for the model, the context and every user at once. Check the hardware before you rent a server.
More AI brand comparisons
Model list and prices come from OpenRouter’s public model catalogue, checked 13 November 2026. Prices set by the developer can change without notice; confirm on the provider’s pricing page before you commit.