Sarvam 1 V0.5 is an open-source language model by sarvamai. Features: 2b LLM, VRAM: 5.1GB, Context: 8K, License: other, LLM Explorer Score: 0.16.
Model Card on HF 🤗: https://huggingface.co/sarvamai/sarvam-1-v0.5
Sarvam 1 V0.5 Benchmarks
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
What is the LLM Explorer Rank (Score)
Sarvam 1 V0.5 Parameters and Internals
| LLM Name | Sarvam 1 V0.5 |
| Repository 🤗 | https://huggingface.co/sarvamai/sarvam-1-v0.5 |
| Model Size | 2b |
| Required VRAM | 5.1 GB |
| Updated | 2026-07-11 |
| Maintainer | sarvamai |
| Model Type | llama |
| Model Files | |
| Model Architecture | LlamaForCausalLM |
| License | other |
| Context Length | 8192 |
| Model Max Length | 8192 |
| Transformers Version | 4.44.0 |
| Tokenizer Class | LlamaTokenizer |
| Vocabulary Size | 64128 |
| Torch Data Type | bfloat16 |
Best Alternatives to Sarvam 1 V0.5
Best Alternatives |
Context / RAM |
Downloads |
Likes |
|---|---|---|---|
| Llama 2B Hf 32768 Fpf | 32K / 3.8 GB | 9 | 1 |
| ...icpm 2B Sft Bf16 Llamafied 16K | 16K / 6 GB | 560 | 1 |
| Aitana 2B S IP Instruct | 8K / 4.5 GB | 67 | 0 |
| Aitana 2B S Instruct Aligned | 8K / 6.9 GB | 100 | 0 |
| Aitana 2B SI Instruct Aligned | 8K / 4.5 GB | 39 | 0 |
| Aitana 2B S Base IP 1.0 | 8K / 4.5 GB | 186 | 0 |
| Aitana 2B SI Instruct | 8K / 4.5 GB | 22 | 0 |
| Salamandra 2B Fft Trenes | 8K / 4.5 GB | 32 | 0 |
| SmolLM2 MedIT Upscale 2B | 8K / 4.2 GB | 13 | 5 |
| Salamandra 2B Instruct | 8K / 4.5 GB | 5351 | 28 |
Note: green Score (e.g. "73.2") means that the model is better than sarvamai/sarvam-1-v0.5 .
Rank the Sarvam 1 V0.5 Capabilities
🆘 Have you tried this model? Rate its performance. This feedback would greatly assist ML community in identifying the most suitable model for their needs. Your contribution really does make a difference! 🌟
What open-source LLMs or SLMs are you in search of? 54964 in total.
Original data from HuggingFace, OpenCompass and various public git repos.
Check out Ag3ntum — our secure, self-hosted AI agent for server management.
Release v20260328a