Sarvam
Sarvam 30B (high)
Quality Score7.0
Value Score9.1
Reliability14.2
Cache Discount38%
Model specification
- Reasoning
- reasoning
- Input modalities
- text
- Output modalities
- text
- Context window
- 65,536 tokens
- Weights
- open weights
Token prices USD per 1M tokens
- Input
- $0.026
- Output
- $0.110
- Cached Input
- $0.016
- Cached Output
- not available
- Cost per task
- not available
Capability
- Intelligence
- 6.4
- Coding
- not available
- Agentic
- not available
- Omniscience
- -71.5
- Correct
- 12.6%
- Blended price
- $0.027
Answer outcomes
Correct 12.6%Incorrect 84.1%Abstained 3.2%
Artificial Analysis benchmarks
- SciCode
- 19%
- Humanity’s Last Exam
- 8%
- GPQA Diamond
- 63%
- CritPt
- 0%
- AA-Omniscience
- 14%
- AA-LCR
- 0%
Similar models
- Sarvam 105B (high)SarvamQuality Score 12.4Value Score 15.2
- Qwen3.5 2B (Reasoning)AlibabaQuality Score 6.5Value Score not available
- NVIDIA Nemotron 3 Nano 4BNVIDIAQuality Score 7.4Value Score not available
- Reka Flash 3Reka AIQuality Score 6.8Value Score 6.1
- Olmo 3 7B ThinkAllen Institute for AIQuality Score 5.3Value Score not available
- Qwen3.5 0.8B (Reasoning)AlibabaQuality Score 4.0Value Score not available