DeepSeek V4.1 Flash
DeepSeek · 2 thinking configurations
Intelligence39.5
Input / 1M$0.30
Output / 1M$1.20
Context1M
Tariff: DeepSeek · Source: artificial-analysis · USD per million tokens
Thinking configuration
What does more thinking buy you?
Compare this model’s measured thinking configurations.
Scroll the chart ↔ · tap a point for details
Efficient frontier: best measured score at each cost2 measured configurations · 0 without both values
Selected: thinking max · Click a point to switch configuration
DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture. It activates 8B parameters on input and 16B on...
Intelligence IndexArtificial Analysis composite index, not a percentage. Different versions of the index are not directly comparable.
39.5 pointsHumanity’s Last ExamDifficult questions across many disciplines, compared under the same evaluation setup.
39.2 %SciCodeCode generation for scientific research problems.
51.9 %MMMU-ProMultimodal understanding of academic problems that include images.
77.0 %Omniscience accuracyFactual accuracy in Artificial Analysis's Omniscience evaluation.
46.4 %Specifications from the catalog
Inputtext, image
Outputtext
Tool callingYes
Structured outputYes
Speed209.4 tok/s
First token0.94 s
Maximum output943.7K
Catalog listing date2026-09-10