Open weights · DeepSeek Model License
DeepSeek V3
DeepSeek's 671B MoE general-purpose model (37B active), an open-weight option for coding and reasoning.
Key Specifications
- Context window
- 128K tokens
- Released
- Parameters
- 671B, 37B active
- Inputs
- Text
- Capabilities Show details
- Function calling Connect to external tools, APIs, and systems.
Hosted API pricing
| Provider | Input /1M tokens | Output /1M tokens | Cached input /1M | Cost at 10M in + 2M out | |
|---|---|---|---|---|---|
Novita
Cheapest
|
$0.27 | $1.12 | $0.14 | $4.94 | View |
Azure
|
$1.14 | $4.56 | $20.52 | View | |
Replicate
|
$1.45 | $1.45 | $17.40 | View |
Heads up: Base-tier, on-demand rates per 1M tokens; cached, batch and long-context tiers excluded. A provider may serve a shorter context or a quantized build than the creator's release. Verify before provisioning. More on how we price.
Estimated cost to self-host
DeepSeek V3 needs about 373 GB of GPU memory at 4-bit with 32K context. The cheapest rental that fits is 8× A100 at about $10,138 a month.
That costs the same as roughly 25B tokens a month on Novita's API. Self-hosting is more expensive below that volume.
| Precision | Memory | Cheapest fit (32K context) | Cost /mo | Break-even vs API |
|---|---|---|---|---|
| 4-bitINT4 / FP4 |
373 GB
|
$10,138
|
25B tokens /mo
|
|
| 8-bitFP8 / INT8 |
674 GB
|
$12,672
|
31B tokens /mo
|
|
| 16-bitFP16 / BF16 |
1,346 GB
|
$45,446
|
110B tokens /mo
|
Estimates based on median on-demand rates for Nvidia GPUs. Memory is weights plus KV cache for one request, using FP8 KV cache where supported and FP16/BF16 otherwise. Break-even assumes a 5:1 input-to-output ratio. No guarantee of runtime support, usable performance, or that a matching quantized build exists. How we estimate costs.
Similarly priced models
The models nearest DeepSeek V3 by blended rate, each at its own cheapest provider.
| Model | Blended /1M | Input /1M | Output /1M | Context | Cutoff | vs DeepSeek V3 |
|---|---|---|---|---|---|---|
GPT-5.4 Nano
OpenAI
|
$0.375 | $0.20 | $1.25 | 400K | Aug 2025 | −9% |
Qwen2.5-72B Instruct
Alibaba Cloud
|
$0.3833 | $0.38 | $0.40 | 33K | −7% | |
|
DeepSeek V3.1
DeepSeek
|
$0.3917 | $0.27 | $1.00 | 131K | −5% | |
|
DeepSeek V3.1 Terminus
DeepSeek
|
$0.3917 | $0.27 | $1.00 | 131K | −5% | |
Codestral
Mistral
|
$0.40 | $0.30 | $0.90 | 128K | −3% | |
|
DeepSeek V3
This model
DeepSeek
|
$0.4117 | $0.27 | $1.12 | 128K | ||
|
Qwen3.7-Plus
Alibaba Cloud
|
$0.4167 | $0.28 | $1.10 | 1M | +1% | |
|
Qwen3-VL-Plus
Alibaba Cloud
|
$0.4333 | $0.20 | $1.60 | 262K | +5% | |
|
DeepSeek V4.1 Flash
DeepSeek
|
$0.45 | $0.30 | $1.20 | 1M | +9% | |
MiniMax-M2.5
MiniMax
|
$0.45 | $0.30 | $1.20 | 205K | +9% | |
Gemini 3.1 Flash-Lite
Google Cloud
|
$0.4583 | $0.25 | $1.50 | 1M | Jan 2025 | +11% |
Prices are USD per 1M tokens at each model's cheapest listed provider. Blended is the cost of 10M input plus 2M output tokens, spread over the 12M.
Frequently Asked Questions
What is DeepSeek V3 good for?
General chat and coding with tool calling. Open weights under the DeepSeek Model License, on a multi-GPU node.
When is DeepSeek V3 not a good fit?
Image input or structured output: it is text only. Its weights carry the DeepSeek Model License rather than the MIT terms of the later releases, and running them takes a multi-GPU node.
What is the cheapest way to run DeepSeek V3?
Hosted, unless you push serious volume. Novita charges $0.27 in / $1.12 out per 1M tokens. The cheapest rental that fits is 8x A100 at $10,138 a month, which costs the same as about 25B tokens a month on that API.
How does DeepSeek V3 compare with DeepSeek V3.1?
DeepSeek says DeepSeek V3.1 continues pretraining from DeepSeek V3 to extend its context, then post-trains one model with thinking and non-thinking modes, with better tool use and multi-step agent tasks. Same text input. DeepSeek V3 costs 12% more per 1M output tokens than DeepSeek V3.1, at each model's cheapest listed provider. It came out 8 months earlier, has a smaller context (128K against 131K tokens) and is licensed under DeepSeek Model License against MIT.
Can I self-host DeepSeek V3?
Yes. The weights are released under the DeepSeek Model License. At 4-bit it needs about 373 GB of GPU memory, which starts at roughly $10,138 a month on the cheapest rental that fits.
More from DeepSeek
| Model | Context | Input /1M | Output /1M |
|---|---|---|---|
|
DeepSeek V3.1
Open weights
|
131K | $0.27 | $1.00 |
|
DeepSeek V3.1 Terminus
Open weights
|
131K | $0.27 | $1.00 |
|
DeepSeek V4.1 Flash
Open weights
|
1M | $0.30 | $1.20 |
|
DeepSeek V3.2
Open weights
|
131K | $0.25 | $0.80 |
|
DeepSeek V4 Pro
Open weights
|
1M | $0.45 | $0.89 |
|
DeepSeek V3 Turbo
Open weights
|
64K | $0.40 | $1.30 |
|
DeepSeek V4 Flash Vision Exp
|
1M | $0.44 | $1.32 |
|
DeepSeek R1 Distill Llama 70B
Open weights
|
131K | $0.80 | $0.80 |