← All models

DeepSeek R1 0528

DeepSeek

reasoning

Capabilities

ToolsCodeExtended ThinkingStructured Output

Strengths

  • Major upgrade over R1 — better instruction following
  • Fewer reasoning loops
  • Agentic tool use
  • MIT licensed

Weaknesses

  • No vision
  • Higher output cost (chain-of-thought)
  • Data jurisdiction

Pricing

Input / 1M tokens

$0.5

Output / 1M tokens

$2.1500000000000004

Context window

128k

Transparency

Open weights

10.0 / 10

Open training data

3.0 / 10

Open methodology

7.0 / 10

Licence openness

10.0 / 10

Provider disclosure

2.0 / 10

FMTI company score

300.0 / 10

Composite:6.4 / 10

MIT licence + open weights. DeepSeek FMTI 30.

Sustainability

Inference energy

3.5 / 10

Training footprint

N/A

Provider infrastructure

2.0 / 10

Composite:2.8 / 10

Reasoning = more tokens = higher energy.

Task Fitness

question answering
82%
code
90%
mathematics
91%
business communication
61%
analysis
92%
extraction
78%
long-form generation
72%
research
64%
reasoning
92%
summarisation
75%
translation
68%
conversation
68%