DeepSeek V3.1 is an open-source chat model created by DeepSeek. It is a hybrid model designed for tasks like general conversation, web search, code execution, and function calling.
Input
Output
Context
131K
Max Output
-
Parameters
684.5B
Input Modalities
Output Modalities
Some values come from web research and may be incorrect. Report incorrect data
Bands compare this model with every model the same suite measured in our catalog.
Download 689 GB · Runs on vLLM v0.32.0 · 9 GGUF builds
| Precision | Weights | One request at 33K tokens | Smallest GPU that fits |
|---|---|---|---|
| FP16 / BF16 | 1275 GB | 1357 GB | 4x AMD Instinct MI455X (1728 GB) |
| INT8 | 638 GB | 688 GB | 2x AMD Instinct MI455X (864 GB) |
| INT4 | 319 GB | 353 GB | 1x AMD Instinct MI455X (432 GB) |
Computed by Inferbase from the 684,531,386,000 parameters in the creator's Hugging Face repository; one request adds its KV cache, activations and framework overhead, with the layer layout estimated from the parameter count.
Request deepseek-v3-1 and routing picks a build. Use a build's model ID to pin it.
| Build | What it is | Model ID |
|---|---|---|
| FP4 | 4-bit floating-point weights | deepseek-v3-1:fp4 |
Public resources
Specification sources
As of Nov 10, 2026.
Answered from the facts on this page, with their sources.
DeepSeek V3.1 has a context window of 131,072 tokens (source: Inferbase review). A provider may serve less; Inferbase routes on what each provider serves.
Yes. DeepSeek V3.1 is an open-weight model under the MIT license. You can download and self-host it within the license terms.
DeepSeek V3.1 has 684,531,386,000 parameters (684.5B), as stated in the creator's Hugging Face repository.
Strengths: Open weights under a permissive license (mit); In the top 25% on knowledge (MMLU-Pro) among measured models. Limitations: Below average on long-context reasoning among measured models. Each is computed from the model's stated facts and benchmark results.
DeepSeek V3.1 accepts Text and produces Text. The Capabilities card lists the API features it supports, such as function calling and structured output.
DeepSeek V3.1 has published results from Artificial Analysis and Epoch AI. The Benchmarks card shows each score as reported, with its band among the models the same source measured. Scores are never combined across sources.
Yes. DeepSeek V3.1 is available on Inferbase Managed through the OpenAI-compatible API as deepseek-v3-1. Request it by name, or request auto and routing picks it when it fits the request best.
The weights take 1.28 TB at FP16 / BF16, 637.5 GB at INT8, 318.8 GB at INT4, computed from the 684,531,386,000 parameters its creator publishes. Each request adds its KV cache and the framework's overhead. The hardware section shows one request and the smallest GPU that fits; the GPU planner sizes it for your context length and traffic.
Start on the free plan with 5,000 requests a month, no card required.