DEV.co
Open-Source LLM · sarvamai

sarvam-30b

Sarvam-30B is a 30-billion-parameter open-source Mixture-of-Experts (MoE) language model optimized for multilingual support across 22 Indian languages and efficient deployment in resource-constrained environments. With only 2.4B active parameters, it balances reasoning, coding, and conversational capability while maintaining lower memory footprint than dense models of similar scale. Licensed under Apache 2.0 with no gating restrictions, it is suitable for self-hosted and custom LLM applications.

Source: HuggingFace — huggingface.co/sarvamai/sarvam-30b
32.2B
Parameters
apache-2.0
License (OSI-approved)
Unknown
Context (tokens)
55.3k
Downloads (30d)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

FieldValue
Developersarvamai
Parameters32.2B
Context windowUnknown
Licenseapache-2.0 — OSI-approved
Modality / tasktext-generation
Gated on HuggingFaceNo
Downloads55.3k
Likes211
Last updated2026-03-23
Sourcesarvamai/sarvam-30b

What sarvam-30b is

Sarvam-30B uses a 19-layer MoE architecture with 128 routed experts plus one shared expert, top-6 routing, grouped KV attention (4 heads), and dense FFN intermediate size 8192 with MoE intermediate size 1024. It employs high rope_theta (8e6) for long-context stability without RoPE scaling, auxiliary-loss-free router balancing, and 2.5x routed scaling factor. The model supports context lengths up to 65,536 tokens in benchmarking. Evaluated with transformers, safetensors, and custom_code=true, requiring trust_remote_code. Inference backends include HuggingFace, vLLM (via hot-patch or source fork), and SGLang.

Quickstart

Run sarvam-30b locally

Load the open weights with 🤗 Transformers and generate — the same model, self-hosted.

quickstart.pypython
from transformers import pipelinepipe = pipeline("text-generation", model="sarvamai/sarvam-30b")out = pipe("Explain retrieval-augmented generation in one sentence.",           max_new_tokens=128)print(out[0]["generated_text"])

Swap in vLLM or Ollama for production-grade serving. DEV.co can stand up the inference stack.

Deployment

How you'd run it

A typical self-hosted path — open weights, an inference server, your application.

DEV.co builds each layer — from GPU infrastructure to the application.

Best use cases

Indian Language NLP Applications

Sarvam-30B achieves state-of-the-art performance across 22 Indian languages for its size, making it ideal for customer support, content generation, and conversational AI targeting Indian markets without dependency on closed-source APIs.

Resource-Constrained Self-Hosted Deployments

With 2.4B active parameters via MoE routing, the model runs efficiently on smaller GPU clusters compared to dense 30B alternatives. Suitable for on-premise deployment where commercial model APIs are unavailable or unacceptable.

Agentic and Tool-Calling Workflows

Demonstrated tool-calling capability (AIME 25: 96.7% with tools) and agentic benchmarks (BrowseComp, SWE Bench) make it suitable for autonomous task agents, code generation pipelines, and browser-based automation.

Running & fine-tuning it

ESTIMATE (requires independent verification): 30B parameters with MoE architecture and 2.4B active parameters. In bfloat16 precision (~60 GB disk storage), activation-only inference approximately 5–12 GB VRAM per GPU depending on batch size and routing sparsity. Multi-GPU inference with tensor parallelism (tp_size=2–8) recommended for throughput. vLLM example uses tp_size=8. Quantized variants (int8, int4) would reduce footprint further but not mentioned in card.

Model card does not explicitly document LoRA, QLoRA, or full fine-tuning support. Custom code required (trust_remote_code=True). Standard HuggingFace transformer-based LoRA should be feasible given transformers framework compatibility, but adapter compatibility and training stability on MoE architecture not documented. Recommend testing on small dataset before production fine-tuning.

When to avoid it — and what to weigh

  • Strict Long-Context Requirements Beyond 65K Tokens — Benchmark evaluation maxes at 65,536 tokens. Production context length limits are not stated. If applications require 100K+ token context windows with proven performance, requires independent testing.
  • Minimal Agentic Performance Needed — Agentic benchmarks show mixed results: BrowseComp (35.5%), SWE Bench (34.0%), τ² Bench (45.7%) trail some competing models (e.g., GLM 4.7 Flash at 79.5% τ²). For mission-critical agent systems, compare against specialized alternatives.
  • Maximum Inference Latency Sensitivity — MoE routing overhead and sparse expert selection may introduce variable latency. Inference speed benchmarks not provided. Dense alternatives may offer more predictable throughput for real-time applications.
  • Weak Coverage of Non-Indian Language Domains — Model is optimized for Indian languages and English. Performance on other non-English languages not clearly documented. Not recommended as primary choice for Japanese, Chinese, or Arabic-heavy workloads.

License & commercial use

Apache License 2.0 (SPDX: apache-2.0). Permissive OSI-approved open-source license. No gating or model agreement required (gated=false). Permits commercial use, modification, and distribution with attribution and Apache 2.0 boilerplate.

Apache 2.0 is a permissive open-source license that explicitly permits commercial use and distribution. No royalty obligations, proprietary license fees, or commercial restrictions apply. Users may deploy, modify, and resell applications incorporating Sarvam-30B. However, liability disclaimers and patent grant limitations in Apache 2.0 should be reviewed by legal counsel for high-risk deployments. No warranty or indemnification provided by sarvamai.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

SignalAssessment
MaintenanceActive
DocumentationAdequate
License clarityClear
Deployment complexityModerate
DEV.co fitStrong
Assessment confidenceHigh
Security considerations

Model requires trust_remote_code=True for HuggingFace and vLLM loading, introducing code execution risk if model repository is compromised or tampered. Standard LLM security considerations apply: no built-in guardrails or safety filters documented; evaluate output for bias, toxicity, and hallucinations before production deployment, especially in sensitive Indian language contexts. No security audit or adversarial robustness testing mentioned. No rate-limiting, input sanitization, or prompt injection mitigations documented.

Alternatives to consider

Nemotron-3-Nano-30B (NVIDIA)

Similar 30B parameter class, stronger agentic benchmarks (SWE Bench 38.8%, τ² Bench 49.0%), comparable coding performance. Closed-source but NVIDIA-backed. Choose if agentic capability is priority over Indian language support.

Qwen3-30B-Thinking (Alibaba)

30B dense model with explicit reasoning/thinking mode, superior MMLU Pro (80.9%) and math (Math500 97.6%), strong agentic benchmarks (τ² 47.7%). Better for pure reasoning workloads; MoE trade-off may not be necessary.

Mistral 3.2-24B

Smaller dense model, permissive license, strong HumanEval (92.9%), well-documented. Lower resource footprint but weaker knowledge benchmarks (MMLU 80.5%, MMLU Pro 69.1%) and no Indian language specialization.

Software development agency

Ship sarvam-30b with senior software developers

Sarvam-30B offers permissive Apache 2.0 licensing and strong Indian language support for self-hosted applications. Evaluate memory and latency requirements on your hardware, test fine-tuning if needed, and integrate via HuggingFace, vLLM, or SGLang. Contact Devco to architect your deployment.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

sarvam-30b FAQ

Can I use Sarvam-30B in a commercial product without paying licensing fees?
Yes. Apache 2.0 permits commercial use without royalties or fees. You must retain copyright notices and include a copy of the Apache 2.0 license. No warranty is provided by sarvamai; review Apache 2.0 liability and patent terms with legal counsel for production systems.
What GPU memory is needed to run Sarvam-30B inference?
Estimated 5–12 GB VRAM in bfloat16 depending on batch size and routing sparsity. The model card does not specify exact memory profiles. Multi-GPU tensor parallelism (tp_size=2–8) is recommended for production throughput. Independent benchmarking on your target hardware is required.
Can I fine-tune Sarvam-30B on my own data?
Likely feasible via LoRA or full fine-tuning using HuggingFace transformers, but not explicitly documented. The model uses custom_code=True, requiring trust_remote_code. Test fine-tuning on a small dataset first; MoE-specific training stability is not discussed in the card.
Does Sarvam-30B support languages outside India?
Primary optimization is for English and 22 Indian languages. Performance on other major languages (Chinese, Japanese, Spanish, French) is not documented. Not recommended as primary model for non-Indian language workloads.

Work with a software development agency

DEV.co is a software development agency delivering custom software development services to companies building on open source. Our software developers and web developers design, integrate, and ship production systems — spanning web development, APIs, AI, data, and cloud. If sarvam-30b is part of your open-source llms roadmap, our team can implement, customize, migrate, and maintain it.

Ready to Deploy Multilingual AI in Your Infrastructure?

Sarvam-30B offers permissive Apache 2.0 licensing and strong Indian language support for self-hosted applications. Evaluate memory and latency requirements on your hardware, test fine-tuning if needed, and integrate via HuggingFace, vLLM, or SGLang. Contact Devco to architect your deployment.