DeepSeek V3.1
Summary
DeepSeek V3.1 is an open-weight language model from DeepSeek, released on 21 Aug 2025. It accepts text and generates text, with a 164K-token context window and up to 33K output tokens per response. As of 17 Oct 2026, the lowest listed API price is $0.25 per million input tokens and $0.95 per million output tokens (OpenRouter), across 1 provider we track.
DeepSeek chat model for instruction following, coding, and analysis
Last verified 17 Oct 2026. Every figure on this page links to its source.
Specifications
| Developer | DeepSeek |
|---|---|
| Release date | 21 Aug 2025 [models.dev] |
| Status | Generally available [models.dev] |
| Weights | Open weights [Hugging Face] |
| License | mit [Hugging Face] |
| Parameters | 685B [Hugging Face] |
| Context window | 164K (163,840 tokens) [models.dev] |
| Max output | 33K tokens [models.dev] |
| Knowledge cutoff | 31 Mar 2025 [models.dev] |
| Input | Text [models.dev] |
| Output | Text [models.dev] |
| Reasoning mode | Yes [models.dev] |
| Tool calling | Yes [models.dev] |
| Structured output | Yes [models.dev] |
| Hugging Face | deepseek-ai/DeepSeek-V3.1 [Hugging Face] |
DeepSeek V3.1 API pricing by provider
USD per million tokens. Sorted by input price.
| Provider | Input | Output | Cached input | Context | Source |
|---|---|---|---|---|---|
| OpenRouter deepseek/deepseek-chat-v3.1 | $0.25 | $0.95 | $0.13 | 164K | OpenRouter, 17 Oct 2026 |
Running DeepSeek V3.1 locally
Estimated memory, from the parameter count and the architecture in config.json on Hugging Face. Includes about 0.8 GB of runtime overhead.
| Quantization | Weights | Total, 8K context | Total, 32K context |
|---|---|---|---|
| Q4_K_M (4-bit) | 415 GB | 416 GB | 418 GB |
| Q8_0 (8-bit) | 727 GB | 729 GB | 730 GB |
| FP16 / BF16 (16-bit) | 1369 GB | 1370 GB | 1372 GB |
At 4-bit with an 8K conversation, DeepSeek V3.1 needs about 416 GB, more than any single consumer graphics card; it needs a unified-memory machine, a multi-GPU setup or a split between graphics card and system RAM. Check your own hardware.
Frequently asked questions
What is the context window of DeepSeek V3.1?
DeepSeek V3.1 has a context window of 163,840 tokens (164K) and can generate up to 32,768 tokens in a single response, according to models.dev.
How much does the DeepSeek V3.1 API cost?
DeepSeek V3.1 is listed from $0.25 per million input tokens and $0.95 per million output tokens (OpenRouter). Prices last verified 17 Oct 2026.
Is DeepSeek V3.1 open source?
DeepSeek V3.1 is an open-weight model: its weights are published on Hugging Face as deepseek-ai/DeepSeek-V3.1 under the mit license. Check the license terms before commercial use.
When was DeepSeek V3.1 released?
DeepSeek released DeepSeek V3.1 on 21 Aug 2025, according to models.dev.
How much VRAM does DeepSeek V3.1 need?
About 416 GB at 4-bit (Q4_K_M) with an 8K-token conversation, and about 729 GB at 8-bit. That is more than any single consumer graphics card holds. These are estimates from the model's published parameter count and architecture.
Change history
- License changed from none to mit
- Parameter count changed from none to 685B
- Now listed on OpenRouter
- Added to the directory