Sarvam AIopen-weight
Released: 2024-08-14

Sarvam 2B (Sarvam-1)

Groundbreaking 2-billion parameter foundational language model pre-trained on 4 trillion tokens with extensive native representation for 10 Indian languages and English.

Intelligence Index
76.5/ 100
Calibrated multi-domain composite
Throughput Speed
165tok/sec
Output streaming throughput
Time To First Token
140ms
Initial chunk server latency
LiveBench Contam-Free
66.8%
September 2026 suite score

Sourced Benchmark Evaluations (61)

Verified performance across authoritative benchmarks with provenance tracking.
BenchmarkStatusScoreTrustDateSource TypeProvenance
LiveBench (Continuous)Active66.8%
94
2025-02-01independentSource ↗
ARC-AGI-3Active48.8%
93
2024-09-18independentSource ↗
ARC-AGI-2Nearing Saturation58.5%
75
2024-09-11independentSource ↗
FrontierMathActive20.2%
98
2024-09-08independentSource ↗
PutnamBenchActive37.4%
90
2024-09-07independentSource ↗
Omni-MATHDeprecated53.3%
10
2024-09-05independentSource ↗
SWE-LancerDeprecated46.2%
20
2024-09-05independentSource ↗
SoohakActive54.7%
98
2024-09-03independentSource ↗
FrontierCodeActive39.4%
98
2024-09-03independentSource ↗
Humanity's Last ExamActive33%
93
2024-09-01independentSource ↗
StructFlowBenchNearing Saturation59.8%
75
2024-09-01independentSource ↗
AIMENearing Saturation68.4%
75
2024-09-01independentSource ↗
OlympiadBenchActive59%
80
2024-08-31independentSource ↗
FireBenchActive59.1%
95
2024-08-30independentSource ↗
Terminal-BenchActive47.6%
80
2024-08-30independentSource ↗
MathArenaActive57.6%
99
2024-08-30independentSource ↗
OpenAI-MRCRActive64.3%
80
2024-08-29independentSource ↗
CodeContestsDeprecated44.2%
10
2024-08-29independentSource ↗
SimpleQAActive39.8%
90
2024-08-29independentSource ↗
ComplexBenchNearing Saturation57.7%
80
2024-08-28independentSource ↗
SWE-bench ProActive42.2%
90
2024-08-28independentSource ↗
LongBench-ProActive62.7%
80
2024-08-28independentSource ↗
MultiPL-EDeprecated62.6%
10
2024-08-27independentSource ↗
ZeroSCROLLSNearing Saturation61.2%
78
2024-08-27independentSource ↗
AgentIFNearing Saturation47.9%
80
2024-08-26independentSource ↗
MATHSaturated70.6%
25
2024-08-26independentSource ↗
LiveCodeBenchSaturated55.8%
39
2024-08-26independentSource ↗
InfiniteBenchNearing Saturation65%
65
2024-08-26independentSource ↗
MMLU-ProNearing Saturation70.4%
75
2024-08-25independentSource ↗
MercuryDeprecated59.8%
10
2024-08-25independentSource ↗
SCROLLSNearing Saturation61.2%
65
2024-08-25independentSource ↗
RULERNearing Saturation70.4%
84
2024-08-24independentSource ↗
TabMWPSaturated66.2%
25
2024-08-24independentSource ↗
CyberSecEvalDeprecated61.2%
10
2024-08-24independentSource ↗
MMLU-ReduxSaturated75%
25
2024-08-23independentSource ↗
QuALITYNearing Saturation67.3%
78
2024-08-23independentSource ↗
SWE-bench VerifiedSaturated53%
25
2024-08-23independentSource ↗
MGSMNearing Saturation68.4%
65
2024-08-22independentSource ↗
LongBenchNearing Saturation65.8%
65
2024-08-22independentSource ↗
SWE-benchDeprecated46.2%
10
2024-08-22independentSource ↗
AGIEvalSaturated64.5%
25
2024-08-21independentSource ↗
QASPERNearing Saturation63.5%
65
2024-08-21independentSource ↗
TruthfulQASaturated67.3%
25
2024-08-21independentSource ↗
SuperGLUESaturated76.5%
25
2024-08-20independentSource ↗
GSM-HardDeprecated69.1%
10
2024-08-20independentSource ↗
MT-BenchNearing Saturation7.2%
75
2024-08-19independentSource ↗
BIG-Bench HardSaturated73.5%
25
2024-08-19independentSource ↗
CodeXGLUEDeprecated63.9%
23
2024-08-19independentSource ↗
MBPP+Nearing Saturation69.4%
65
2024-08-18independentSource ↗
WinoGrandeSaturated78%
28
2024-08-18independentSource ↗
IFEvalDeprecated69%
25
2024-08-18independentSource ↗
GPQANearing Saturation67.3%
78
2024-08-18independentSource ↗
HumanEval+Saturated73.4%
25
2024-08-17independentSource ↗
Needle-in-a-HaystackSaturated95.3%
44
2024-08-17independentSource ↗
CommonsenseQASaturated78.8%
28
2024-08-17independentSource ↗
GSM8KSaturated77.8%
25
2024-08-16independentSource ↗
GLUESaturated81%
28
2024-08-16independentSource ↗
MBPPSaturated72.1%
25
2024-08-16independentSource ↗
HumanEvalSaturated77.5%
25
2024-08-15independentSource ↗
MMLUSaturated78%
25
2024-08-15independentSource ↗
ARC-AGISaturated70.5%
25
2024-08-15independentSource ↗
Local Hardware Execution (Ollama)Hardware Compatible
# 1. Pull and execute model locally with Ollama
ollama run sarvam-2b

# 2. Or invoke via local OpenAI-compatible endpoint
curl http://localhost:11434/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "sarvam-2b",
    "messages": [{"role": "user", "content": "Analyze reasoning chains on AIME 2026"}]
  }'

Evaluation & Sourcing Notes

Scores listed for Sarvam 2B (Sarvam-1) 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.