Skip to content
RankShark

DeepSeek V3.1

DeepSeek-V3.1 is a large hybrid reasoning model (671B parameters, 37B active) that supports both thinking and non-thinking modes via prompt

ModelOpen sourceby DeepSeeksince 2025
32.9

SharkScore

#238 in Models

Why it scores 32.9

DeepSeek V3.1 has a SharkScore of 32.9/100, ranked #238 of 608 in Models (updated 2026-11-19).

Each part compares DeepSeek V3.1 with other listings in Models. Evidence covers 87% of the score's weight.

Adoption
25 · 30%
User reviews
no data · 25%
Maintenance
1 · 20%
Trust & transparency
80 · 15%
Momentum
no data · 10%

Paying never changes this score. How ranking works · method v2026-11-18.2

RankShark review of DeepSeek V3.1

Automated evaluation by RankShark · 18 Nov 2026 · method v1.0 · How we review

We were not paid for this review.

As of 2026-11-18, DeepSeek V3.1 is a text-only hybrid reasoning model; OpenRouter lists a 163,840-token context and rates of $0.25/M input and $0.95/M output. [E1, E2, MEASURED]

DeepSeek V3.1 is a 671B-parameter model with 37B active parameters. It supports thinking and non-thinking modes, and the model card describes tool calling and agent use. OpenRouter lists it as text input to text output. [E1, E2]

Best for

  • Applications that need to choose between thinking and non-thinking modes. · model card
  • Text-based tool-calling or agent workflows. · model card, OpenRouter
  • API users who need a context window listed at 163,840 tokens. · OpenRouter

Not for

  • Applications that require image or audio input; OpenRouter lists text as the model's input modality. · OpenRouter

Key features

  • Thinking modes. The vendor says the model supports thinking and non-thinking modes by changing the chat template. · model card
  • Tool calling. The vendor says tool calling is supported in non-thinking mode and describes search-agent use with a user-provided search tool. · model card
  • Long context. The vendor's model table lists a 128K context length; OpenRouter lists 163,840 tokens. · model card, OpenRouter
  • Model size. The vendor lists 671B total parameters and 37B activated parameters. · model card
  • Text modality. OpenRouter lists text input and text output. · OpenRouter

Pros

  • Offers both thinking and non-thinking modes. · model card
  • Supports tool calls, including a documented format for code-agent frameworks. · model card
  • The listed OpenRouter rates are $0.25 per million input tokens and $0.95 per million output tokens. · OpenRouter

Cons

  • OpenRouter lists text-only input and output, not image or audio modalities. · OpenRouter
  • The model card gives specific local-running precision requirements: load and compute one parameter in FP32 and use UE8M0 formatting for FP8 weights and activations. · model card
  • No free plan or free trial stated on the pages we checked.
  • Whether customer data is used for model training is not stated on the pages we checked.

Pricing

PlanPriceLimits
OpenRouter API usage$0.25 per million input tokens; $0.95 per million output tokens; $0.13 per million cached input tokensOpenRouter lists a 163,840-token context window.

Checked 18 Nov 2026

Free plan
Not stated (checked 18 Nov 2026)
Free trial
Not stated (checked 18 Nov 2026)
Trains on customer data
Not stated (checked 18 Nov 2026)
Data retention
Not stated (checked 18 Nov 2026)
Data regions
Not stated (checked 18 Nov 2026)
Data processing agreement
Not stated (checked 18 Nov 2026)
AI models used
DeepSeek-V3.1 · model card, OpenRouter
API
OpenRouter lists chat completions, completions, messages, and responses endpoints for the model. · OpenRouter
SDKs
Not stated (checked 18 Nov 2026)
Webhooks
Not stated (checked 18 Nov 2026)
Platforms
The model card links downloads on Hugging Face and ModelScope; OpenRouter lists the model for API access. · model card, OpenRouter
Security claims
No security certification stated (checked 18 Nov 2026)

What we measured

Context window (OpenRouter listing) · now
163,840 tokens
Price per million tokens (OpenRouter listing) · now
$0.25 in, $0.95 out
Package downloads · 30 days
306,203
How we tested · sources
  • Hourly uptime and response-time checks of the product's URL.
  • Repository, release and download data from GitHub, npm, PyPI and Hugging Face where they exist.
  • We did not sign up, pay, or use the product as a customer.
  • 18 Nov 2026: Plans or prices changed on the pricing page.

Something wrong? Report an error with a public link that shows it; corrections re-run the review.

Top alternatives to DeepSeek V3.1

See all
  • 67.1

    Qwen3.8 27B is an open-weight dense vision-language model from Qwen.

    ModelOpen source
  • 65.5

    Qwen3.6-35B-A3B is an open-weight multimodal model from Alibaba Cloud with 35 billion total parameters and 3 billion active parameters per t

    ModelOpen source
  • 64.7

    Qwen3.8 Flash is a multimodal reasoning model from Alibaba.

    ModelOpen source
  • 64.3

    Qwen3.6 27B is a dense 27-billion-parameter language model from the Qwen Team at Alibaba, released in April 2026.

    ModelOpen source