All models
Text / ReasoningAvailable via API235B total · 22B active

Vikasit 235B MoE

Large MoE flagship. Advanced reasoning, agentic workflows, 262K context.

Overview

Vikasit 235B MoE is a large-scale flagship — 235B total / 22B active — for advanced reasoning, long-context analysis, and agentic workflows. The Apache-licensed workhorse of the MAX tier, served live via the Vikasit API.

Specifications

Total parameters
235B total
Active parameters
22B active
Architecture
Mixture-of-Experts
Context window
262K native, ~1M via YaRN
Modalities
Text in → text out
License
Apache 2.0

Capabilities

  • Advanced reasoning and analysis
  • Agentic, multi-step tool-use
  • 262K native context, ~1M via YaRN
  • Thinking and non-thinking variants
119 languages. Strong English + major Indian languages.

Benchmarks

BenchmarkScore
GPQA-Diamond68.4
AIME 202570.9
LiveCodeBench—
MMLU-ProN/A
MATH-500N/A
HumanEvalN/A

Qwen3-235B-A22B ships multiple checkpoints (Instruct-2507 / Thinking-2507). Confirm the exact served checkpoint's published table; figures shown are indicative of the thinking variant.

Hardware & deployment

PrecisionMemory
bf16~470 GB
INT4~120 GB

Quick start

Call Vikasit 235B MoE through the OpenAI-compatible Vikasit AI API at https://api.vikasit.ai/v1 using the model id vikasit-235b-moe.

OpenAI-compatible Python (Vikasit AI API)
# pip install openai
import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.vikasit.ai/v1",
    api_key=os.environ["VIKASIT_API_KEY"],
)

resp = client.chat.completions.create(
    model="vikasit-235b-moe",
    messages=[
        {"role": "user", "content": "Explain Vikasit 235B MoE in one sentence."}
    ],
)

print(resp.choices[0].message.content)
# or with curl
curl https://api.vikasit.ai/v1/chat/completions \
  -H "Authorization: Bearer $VIKASIT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "vikasit-235b-moe",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Limitations

  • Full parameter set must be in memory (MoE)
  • Per-checkpoint benchmark variance — verify card

Vikasit 235B MoE FAQ

How much does Vikasit 235B MoE cost?

Vikasit 235B MoE is served through the Vikasit AI API on usage-based, pay-as-you-go pricing billed per million input and output tokens — see the Vikasit AI pricing page for current rates. Because it is built on the open-weight Qwen3-235B-A22B (Apache 2.0), you can also self-host the weights for free under the Apache 2.0 licence and pay only for your own compute.

Is Vikasit 235B MoE open weight?

Yes. Vikasit 235B MoE is built on Qwen3-235B-A22B (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 use Vikasit 235B MoE with the OpenAI SDK?

The Vikasit AI API is OpenAI-compatible. Point the OpenAI client's base URL at https://api.vikasit.ai/v1, set your Vikasit API key, and pass "vikasit-235b-moe" as the model. The quick-start snippet above shows the exact Python call.

What context window does Vikasit 235B MoE support?

Vikasit 235B MoE supports a 262K native, ~1M via YaRN context window. It is a 235B total (22B active) Mixture-of-Experts model — full specifications are listed in the table above.

License & attribution

Apache 2.0

Built on Qwen3-235B-A22B (Apache 2.0). Upstream copyright, license, and attribution notices are retained.

Qwen3-235B-A22B model card