All models
Text / ReasoningOpen weights2B total

Vikasit 2B

Edge-optimized. Multilingual, long context, on-device deployment.

Overview

Vikasit 2B is an edge-optimized model on a hybrid-attention MoE base — multilingual, 262K native context, and strong on-device performance. A capable assistant that fits in tight memory budgets.

Specifications

Total parameters
2B total
Architecture
Hybrid MoE (Gated DeltaNet + sparse MoE)
Layers
24
Context window
262K native
Modalities
Text in → text out (multimodal-capable base)
License
Apache 2.0

Capabilities

  • Multilingual on-device assistant
  • 262K native context
  • Strong instruction following for its size
  • Thinking and non-thinking modes
Multilingual. Strong English + major Indian languages.

Benchmarks

BenchmarkScore
MMLU-Pro66.5
GPQA-Diamond45.0
IFEval61.2
AIME 2025N/A
LiveCodeBenchN/A
MATH-500N/A

Built on Qwen3.5-2B; numbers from the Qwen3.5-2B HuggingFace model card.

Hardware & deployment

PrecisionMemory
bf16~4 GB
INT4~1.3 GB

Quick start

Vikasit 2B is an open-weight model. Self-host it with any OpenAI-compatible inference server and call it with the OpenAI SDK as shown below.

OpenAI-compatible Python (self-hosted, e.g. vLLM)
# pip install openai
import os
from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:8000/v1",
    api_key="sk-local",  # self-hosted servers accept any token
)

resp = client.chat.completions.create(
    model="vikasit-2b",
    messages=[
        {"role": "user", "content": "Explain Vikasit 2B in one sentence."}
    ],
)

print(resp.choices[0].message.content)

Limitations

  • Reasoning below 4B+ on hardest tasks
  • Some standard benchmarks not published by base

Vikasit 2B FAQ

How much does Vikasit 2B cost?

Vikasit 2B is an open-weight model built on Qwen3.5-2B (Apache 2.0). Self-hosting the weights is free under the Apache 2.0 licence — you pay only for the hardware or cloud GPUs you run it on. Typical deployment fits the memory profiles listed in the hardware section above.

Is Vikasit 2B open weight?

Yes. Vikasit 2B is built on Qwen3.5-2B (Apache 2.0) and distributed under the Apache 2.0 licence, so the weights are openly available for self-hosting, fine-tuning, and commercial use, subject to the upstream licence terms.

How do I run Vikasit 2B?

Because Vikasit 2B is open weight, you self-host it with any OpenAI-compatible inference server (such as vLLM or SGLang) loaded with the Qwen3.5-2B (Apache 2.0) weights, then call it with the OpenAI SDK by setting the base URL to your own endpoint.

What context window does Vikasit 2B support?

Vikasit 2B supports a 262K native context window. It is a 2B total Hybrid MoE (Gated DeltaNet + sparse MoE) model — full specifications are listed in the table above.

License & attribution

Apache 2.0

Built on Qwen3.5-2B (Apache 2.0). Upstream copyright, license, and attribution notices are retained.