Vikasit 14B
Strong all-rounder. Complex reasoning, long documents, code review.
Overview
Vikasit 14B is a strong all-rounder for complex reasoning, long-document analysis, and code review. The workstation-class dense model with frontier-adjacent quality.
Specifications
- Total parameters
- 14.8B
- Architecture
- Dense transformer
- Layers
- 40
- Attention
- GQA (40 query / 8 KV heads)
- Context window
- 32K native, 131K via YaRN
- Vocabulary
- 151,669
- Modalities
- Text in → text out
- License
- Apache 2.0
Capabilities
- Complex multi-step reasoning
- Long-document analysis and code review
- 131K extended context (YaRN)
- Thinking and non-thinking modes
Benchmarks
| Benchmark | Score |
|---|---|
| MMLU-Pro | 61.0 |
| GPQA-Diamond | 64.0 |
| AIME 2025 | 70.4 |
| MATH-500 | 96.8 |
| LiveCodeBench v5 | 63.5 |
| BFCL v3 | 70.4 |
| IFEval | 85.4 |
| HumanEval | N/A |
Instruct numbers from the Qwen3 Technical Report; MMLU-Pro is the base-model figure. Thinking-mode scores shown.
Hardware & deployment
| Precision | Memory |
|---|---|
| bf16 | ~30 GB |
| INT4 | ~9 GB |
Quick start
Vikasit 14B 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-14b",
messages=[
{"role": "user", "content": "Explain Vikasit 14B in one sentence."}
],
)
print(resp.choices[0].message.content)Limitations
- Higher latency than 8B for interactive use
- MoE models offer better quality-per-compute at scale
Vikasit 14B FAQ
How much does Vikasit 14B cost?
Vikasit 14B is an open-weight model built on Qwen3-14B (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 14B open weight?
Yes. Vikasit 14B is built on Qwen3-14B (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 14B?
Because Vikasit 14B is open weight, you self-host it with any OpenAI-compatible inference server (such as vLLM or SGLang) loaded with the Qwen3-14B (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 14B support?
Vikasit 14B supports a 32K native, 131K via YaRN context window. It is a 14.8B Dense transformer model — full specifications are listed in the table above.
License & attribution
Apache 2.0
Built on Qwen3-14B (Apache 2.0). Upstream copyright, license, and attribution notices are retained.