Back

DeepSeek

DeepSeek R1 (Jan '25)

ReasoningOpen weights
Intelligence
11.4
Rank 388 of 744 measured models
Coding
24.6
Rank 203 of 298 measured models
Output speed
150 t/s
Rank 85 of 403 measured models
Price/1M
$2.50
Rank 483 of 659 measured models
Context window
128K
Rank 773 of 1199 measured models
Cost per task
$0.262
Rank 80 of 193 measured models

DeepSeek R1 (Jan '25) scores 11.4 on the Artificial Analysis Intelligence Index, ranking 388th of 744 measured models. The closest models on this index are K2 Think V2 (11.5), HyperCLOVA X SEED Think (32B) (11.4), and o1-preview (11.4). At $2.50 per million tokens (3:1 input/output blended price), it is cheaper than o1-preview ($28.88).

Artificial Analysis Indices
Intelligence11.4
Rank 388 of 744 measured models
Coding24.6
Rank 203 of 298 measured models
Math68.0
Rank 102 of 270 measured models
Agentic1.1
Rank 153 of 189 measured models
Capability indexes
Performance in specific industries, measured by Artificial Analysis
Finance & Accounting11.2
Rank 178 of 218 measured models
Strategy & Ops9.0
Rank 175 of 218 measured models
Legal12.7
Rank 183 of 218 measured models
Engineering14.9
Rank 184 of 217 measured models
Economics16.2
Rank 196 of 225 measured models
Standard Benchmarks
MMLU Pro84.4%
Rank 38 of 349 measured models
GPQA Diamond70.8%
Rank 315 of 619 measured models
HLE8.5%
Rank 364 of 664 measured models
LiveCodeBench61.7%
Rank 111 of 343 measured models
SciCode38.3%
Rank 187 of 231 measured models
MATH-50096.6%
Rank 31 of 202 measured models
AIME 202468.3%
Rank 44 of 194 measured models
AIME 202568.0%
Rank 102 of 270 measured models
Models at a similar level
The closest models on the Intelligence Index, one configuration per family.

Preferences and analytics

Rybbit: visits, performance, clicks, copies, forms and session replay (masked inputs), with your permission. Refusing does not limit the site.

Privacy

Allow navigation analytics?

With your permission, Rybbit measures visits, performance and interactions (clicks, copies, forms), including session replay, to improve BenchSift. Input values are masked in replays. Refusing does not limit the site.

Your choice is stored for 180 days. Change it at any time using “Privacy preferences” in the footer. Withdrawal reloads the page.

No choice stored: Rybbit is disabled.

Privacy