Sarvam AIproprietary
Released: 2025-05-20

Sarvam Vision (Document AI)

Specialized OCR and multimodal visual reasoning system designed to digitize, parse, and structure messy handwritten and printed government records, bills, and legal documents across Indian scripts.

Intelligence Index
83.5/ 100
Calibrated multi-domain composite
Throughput Speed
88tok/sec
Output streaming throughput
Time To First Token
320ms
Initial chunk server latency
LiveBench Contam-Free
73.2%
September 2026 suite score

Sourced Benchmark Evaluations (61)

Verified performance across authoritative benchmarks with provenance tracking.
BenchmarkStatusScoreTrustDateSource TypeProvenance
ARC-AGI-3Active52.3%
93
2025-06-24independentSource ↗
ARC-AGI-2Nearing Saturation62.8%
75
2025-06-17independentSource ↗
FrontierMathActive21%
98
2025-06-14independentSource ↗
PutnamBenchActive39%
90
2025-06-13independentSource ↗
SWE-LancerDeprecated49%
20
2025-06-11independentSource ↗
Omni-MATHDeprecated55.5%
10
2025-06-11independentSource ↗
SoohakActive57%
98
2025-06-09independentSource ↗
FrontierCodeActive41.8%
98
2025-06-09independentSource ↗
Humanity's Last ExamActive35.4%
93
2025-06-07independentSource ↗
StructFlowBenchNearing Saturation65.3%
75
2025-06-07independentSource ↗
AIMENearing Saturation71.3%
75
2025-06-07independentSource ↗
OlympiadBenchActive61.5%
80
2025-06-06independentSource ↗
MathArenaActive60%
99
2025-06-05independentSource ↗
FireBenchActive64.5%
95
2025-06-05independentSource ↗
Terminal-BenchActive50.4%
80
2025-06-05independentSource ↗
OpenAI-MRCRActive70.1%
80
2025-06-04independentSource ↗
SimpleQAActive43.4%
90
2025-06-04independentSource ↗
CodeContestsDeprecated46.8%
10
2025-06-04independentSource ↗
ComplexBenchNearing Saturation63%
80
2025-06-03independentSource ↗
SWE-bench ProActive44.6%
90
2025-06-03independentSource ↗
LongBench-ProActive68.5%
80
2025-06-03independentSource ↗
ZeroSCROLLSNearing Saturation66.8%
78
2025-06-02independentSource ↗
MultiPL-EDeprecated66.2%
10
2025-06-02independentSource ↗
LiveCodeBenchSaturated59%
39
2025-06-01independentSource ↗
AgentIFNearing Saturation52.2%
80
2025-06-01independentSource ↗
MATHSaturated73.5%
25
2025-06-01independentSource ↗
InfiniteBenchNearing Saturation71%
65
2025-06-01independentSource ↗
MercuryDeprecated63.4%
10
2025-05-31independentSource ↗
SCROLLSNearing Saturation66.8%
65
2025-05-31independentSource ↗
MMLU-ProNearing Saturation76.8%
75
2025-05-31independentSource ↗
RULERNearing Saturation76.8%
84
2025-05-30independentSource ↗
TabMWPSaturated69%
25
2025-05-30independentSource ↗
CyberSecEvalDeprecated64.8%
10
2025-05-30independentSource ↗
SWE-bench VerifiedSaturated56.2%
25
2025-05-29independentSource ↗
MMLU-ReduxSaturated81.8%
25
2025-05-29independentSource ↗
QuALITYNearing Saturation73.5%
78
2025-05-29independentSource ↗
LongBenchNearing Saturation71.8%
65
2025-05-28independentSource ↗
SWE-benchDeprecated49%
10
2025-05-28independentSource ↗
MGSMNearing Saturation71.3%
65
2025-05-28independentSource ↗
AGIEvalSaturated69.2%
25
2025-05-27independentSource ↗
QASPERNearing Saturation69.3%
65
2025-05-27independentSource ↗
TruthfulQASaturated73.5%
25
2025-05-27independentSource ↗
GSM-HardDeprecated72%
10
2025-05-26independentSource ↗
SuperGLUESaturated82.1%
25
2025-05-26independentSource ↗
CodeXGLUEDeprecated67.7%
23
2025-05-25independentSource ↗
MT-BenchNearing Saturation7.9%
75
2025-05-25independentSource ↗
BIG-Bench HardSaturated78.9%
25
2025-05-25independentSource ↗
WinoGrandeSaturated83.7%
28
2025-05-24independentSource ↗
MBPP+Nearing Saturation73.4%
65
2025-05-24independentSource ↗
IFEvalDeprecated75.3%
25
2025-05-24independentSource ↗
GPQANearing Saturation73.5%
78
2025-05-24independentSource ↗
Needle-in-a-HaystackSaturated96.7%
44
2025-05-23independentSource ↗
HumanEval+Saturated77.8%
25
2025-05-23independentSource ↗
CommonsenseQASaturated84.5%
28
2025-05-23independentSource ↗
MBPPSaturated76.3%
25
2025-05-22independentSource ↗
GSM8KSaturated81%
25
2025-05-22independentSource ↗
GLUESaturated86.9%
28
2025-05-22independentSource ↗
MMLUSaturated85.2%
25
2025-05-21independentSource ↗
HumanEvalSaturated82.1%
25
2025-05-21independentSource ↗
ARC-AGISaturated75.7%
25
2025-05-21independentSource ↗
LiveBench (Continuous)Active73.2%
94
2025-02-01independentSource ↗
OpenAI Compatible API ExecutionAPI Ready
from openai import OpenAI

client = OpenAI()
response = client.chat.completions.create(
    model="sarvam-vision",
    messages=[{"role": "user", "content": "Evaluate multi-step logic problem"}],
    temperature=0.2
)
print(response.choices[0].message.content)

Evaluation & Sourcing Notes

Scores listed for Sarvam Vision (Document AI) represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.

All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.