S
Sarvam 30B (high)
Last updated: Recently
Model Type
Text
Max Context
-
Max Output
-
Parameter Scale
-
Sarvam 30B high by Sarvam: Text model; TTFT 1.209s, 303.4 tok/s.
Input Modality
-
Output Modality
-
Inference Speed
303.387 tokens/s
Latest Release Date
3/6/2026
SDK Ecosystem
-
artificial-analysismanufactureraa-bootstrap
Model Overview Fields
name
Sarvam 30B (high)
Release date
3/6/2026
Performance
303.387 tokens/s
model_creator.name
Sarvam
AA Evaluation Scores
AA Intelligence Index
12.3
AA Coding Index
7.9
AA Math Index
-
MMLU Pro
-
GPQA
0.633
HLE
0.07
LiveCodeBench
-
SciCode
0.192
Math-500
-
AIME
-
AIME 2025
-
IFBench
0.265
LCR
0
TerminalBench Hard
0.023
TAU2
0.345
Output Speed (tokens/s)
303.387 tok/s
TTFT (s)
1.209s
First Answer Token (s)
7.801s