readable AI benchmarks

Mistral

Mistral Medium 3.5

reasoning · open weights · Apr 29, 2026

Quality Score30.2
Value Score24.6
Reliability31.6
Cache Discount90%

Model specification

Reasoning
reasoning
Input modalities
text, image
Output modalities
text
Context window
256k tokens
Weights
open weights

Token prices USD per 1M tokens

Input
$1.50
Output
$7.50
Cached Input
$0.150
Cached Output
not available
Cost per task
$0.464

Capability

Intelligence
30.4
Coding
46.9
Agentic
19.2
Omniscience
-36.8
Correct
24.7%
Blended price
$1.16

Answer outcomes

Correct 24.7%Incorrect 61.5%Abstained 13.9%

Artificial Analysis benchmarks

GDPval-AA v2
22%
τ³-Banking
15%
Terminal-Bench v2.1
51%
SciCode
40%
Humanity’s Last Exam
14%
GPQA Diamond
75%
CritPt
0%
AA-Omniscience
32%
AA-LCR
65%

Similar models