Vikasit 0.8B Writer
Improved writer with refined architecture. Mobile and IoT friendly.
Overview
Vikasit 0.8B Writer is a next-generation tiny model on a hybrid-attention MoE base — refined architecture, 262K native context, and surprisingly strong reasoning for its size. Mobile and IoT friendly.
Specifications
- Total parameters
- 0.8B 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
- Long-context (262K) on tiny hardware
- Refined hybrid-attention efficiency
- Mobile / IoT deployment
- Thinking and non-thinking modes
Benchmarks
| Benchmark | Score |
|---|---|
| MMLU-Pro | 42.3 |
| SuperGPQA | 21.3 |
| BFCL v4 | 25.3 |
| IFEval | 52.1 |
| GPQA-Diamond | N/A |
| AIME 2025 | N/A |
Built on Qwen3.5-0.8B; numbers from the Qwen3.5-0.8B HuggingFace model card. GPQA-Diamond not published (SuperGPQA reported instead).
Hardware & deployment
| Precision | Memory |
|---|---|
| bf16 | ~1.6 GB |
| INT4 | ~0.5 GB |
Quick start
Vikasit 0.8B Writer is an open-weight model. Self-host it with any OpenAI-compatible inference server and call it with the OpenAI SDK as shown below.
# 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-writer-0.8b",
messages=[
{"role": "user", "content": "Explain Vikasit 0.8B Writer in one sentence."}
],
)
print(resp.choices[0].message.content)Limitations
- Tiny model — reasoning depth limited
- Some standard benchmarks not published by base
Vikasit 0.8B Writer FAQ
How much does Vikasit 0.8B Writer cost?
Vikasit 0.8B Writer is an open-weight model built on Qwen3.5-0.8B (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 0.8B Writer open weight?
Yes. Vikasit 0.8B Writer is built on Qwen3.5-0.8B (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 0.8B Writer?
Because Vikasit 0.8B Writer is open weight, you self-host it with any OpenAI-compatible inference server (such as vLLM or SGLang) loaded with the Qwen3.5-0.8B (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 0.8B Writer support?
Vikasit 0.8B Writer supports a 262K native context window. It is a 0.8B total Hybrid MoE (Gated DeltaNet + sparse MoE) model — full specifications are listed in the table above.
License & attribution
Apache 2.0
Built on Qwen3.5-0.8B (Apache 2.0). Upstream copyright, license, and attribution notices are retained.