readable AI benchmarks (simplified)

readable AI benchmarks

DeepSeek

DeepSeek V3.1 Terminus (Reasoning)

reasoning · open weights · Sep 22, 2025

DeepSeek V3.1 Terminus (Reasoning) has a Quality Score of 29.8, ranking 252nd among 551 scored models. Per 1M tokens, pricing is $1.64 input, $2.75 output, and $1.57 cached input. It has a 128k token context window. Its Reliability Score is 36.8, ranking 242nd among 553 scored models, above average (total average: 35.6). Its Value Score is 19.5, low compared with the total Value average of 30.7.

Quality Score29.8coverage 100.0%: AA Intelligence and AA-Omniscience outcomes available
Value Score19.5
Factual reliability36.8
Cache Discount4%

Model specification

Reasoning
reasoning
Input modalities
none
Output modalities
none
Context window
128k tokens
Weights
open weights

Token prices USD per 1M tokens

Input
$1.64
Output
$2.75
Cached Input
$1.57
Cache write
not available
Cost per task
not available

Capability

Intelligence
14.8
Coding
not available
Agentic
not available
Omniscience
-26.4
Correct
27.7%
Blended price
$1.70

Answer outcomes

Fully graded outcomes (Correct + Incorrect): 81.8%

Correct 27.7%Incorrect 54.1%Partial / not attempted 18.2%

Artificial Analysis benchmarks

GDPval-AA v2
10%
τ³-Banking
21%
SciCode
38%
Humanity’s Last Exam
16%
GPQA Diamond
79%
CritPt
2%
AA-Omniscience
37%
AA-LCR
69%

Similar models

Support me! Patreon