All models
31 models. One API.
31 models. One API.
Powered by NVIDIA NIM.
Every Plugsky model runs on NVIDIA NIM — NVIDIA's inference platform hosting 100+ open and proprietary models. Free, paid, vision, reasoning, code, and embeddings. Switch models in a single line.
🆓 Free tier
💎 Paid tier
🧠 Reasoning
👁️ Vision
💻 Code
📚 Long context
🔢 Embeddings
31
Total models
2
Free tier
27
Paid tier
2
Embeddings
1M
Max context
100%
NVIDIA NIM
🆓 Free tier
Available on every account, including free. No rate-limit headroom but perfect for evaluation, learning, and small jobs.
plugsky-micro
Free
nvidia/nvidia-nemotron-nano-9b-v2
Context: 131K tokens
ChatStreamingJSON mode
Fast, cheap — classification, simple chat, intent detection.
model: "plugsky-micro"plugsky-lite
Free
meta/llama-3.2-3b-instruct
Context: 32K tokens
ChatStreamingJSON modeTools
Support & chat automation — moderate complexity.
model: "plugsky-lite"💎 Paid tier — General purpose
The everyday workhorses. Balanced quality, latency, and cost. Available on paid plans (trial / starter / builder / scale / enterprise).
plugsky-plus
Paid
deepseek-ai/deepseek-v4-flash
Context: 32K tokens
ChatStreamingJSONToolsFunction-call
Balanced general agent — good quality at lower cost.
model: "plugsky-plus"plugsky-pro
Paid
qwen3.6-plus → minimaxai/minimax-m3
Context: 65K tokens
ChatStreamingJSONFunction-callCode
Coding & reasoning (default) — strong general purpose.
model: "plugsky-pro"plugsky-max
Paid
nvidia/llama-3.3-nemotron-super-49b-v1.5
Context: 131K tokens
ChatStreamingJSONFunction-callLong-ctx
Complex multi-step — deep reasoning.
model: "plugsky-max"plugsky-frontier
Paid
mistralai/mistral-large-3-675b-instruct-2512
Context: 131K tokens
ChatStreamingJSONFunction-callLong-ctxTools
Frontier-tier — Mistral Large 3 675B (0.3s, 128K context, EU origin).
model: "plugsky-frontier"🧠 Paid tier — Reasoning & code
Models tuned for specific workloads. Higher quality on their target task; same OpenAI-compatible API.
plugsky-reasoning
Paid
nvidia/nemotron-3-super-120b-a12b
Context: 65K tokens
ReasoningToolsJSONLong-ctx
Deep reasoning, math, code — NVIDIA Nemotron Super 120B.
model: "plugsky-reasoning"plugsky-coder
Paid
bytedance/seed-oss-36b-instruct
Context: 131K tokens
CodeToolsJSONLong-ctxMoE
Best open coding model — Qwen3 Coder 480B MoE.
model: "plugsky-coder"plugsky-coder-fast
Paid
meta/llama-3.2-3b-instruct
Context: 32K tokens
CodeToolsJSON
Fast coding — Llama 3.2 3B.
model: "plugsky-coder-fast"plugsky-gpt-oss
Paid
openai/gpt-oss-120b
Context: 32K tokens
ReasoningToolsJSONLong-ctx
Open-source GPT — gpt-oss-120B.
model: "plugsky-gpt-oss"plugsky-phi
Paid
nvidia/nemotron-mini-4b-instruct
Context: 131K tokens
ChatStreamingToolsJSON
NVIDIA Nemotron Mini 4B — ultra-compact, very fast.
model: "plugsky-phi"plugsky-tiny
Paid
nvidia/nvidia-nemotron-nano-9b-v2
Context: 131K tokens
ChatStreamingToolsJSON
NVIDIA Nemotron Nano 9B v2 — small, fast, low cost.
model: "plugsky-tiny"🌊 Paid tier — DeepSeek & Qwen
Open-source powerhouses from China. Top-tier reasoning and code at competitive pricing.
plugsky-deepseek-pro
Paid
deepseek-ai/deepseek-v4-pro
Context: 65K tokens
ReasoningCodeToolsJSON
Reasoning + code — DeepSeek V4 Pro.
model: "plugsky-deepseek-pro"plugsky-deepseek-flash
Paid
deepseek-ai/deepseek-v4-flash
Context: 32K tokens
ChatStreamingToolsJSON
Fast DeepSeek — V4 Flash.
model: "plugsky-deepseek-flash"plugsky-qwen-next
Paid
qwen/qwen3-next-80b-a3b-instruct
Context: 131K tokens
ToolsJSONLong-ctxMoE
Alibaba Qwen3 Next 80B (MoE, 256K context).
model: "plugsky-qwen-next"👁️ Paid tier — Multimodal & long context
Vision, video, ultra-long context, and MoE models. Use these for images, video, and documents that exceed typical 32K context windows.
plugsky-minimax
Paid
minimaxai/minimax-m3
Context: 32K tokens
VisionVideoMultimodalToolsJSON
NVIDIA MiniMax-M3 — strong multimodal + reasoning.
model: "plugsky-minimax"plugsky-ultra
Paid
nvidia/nemotron-3-nano-omni-30b-a3b-reasoning
Context: 1M tokens
VisionMultimodalToolsLong-ctxMoEReasoning
NVIDIA Nemotron 3 Nano Omni 30B — omni-modal + reasoning, 1M context.
model: "plugsky-ultra"plugsky-vision-fast
Paid
meta/llama-3.2-11b-vision-instruct
Context: 32K tokens
VisionMultimodalJSON
Multimodal fast — Llama 3.2 Vision 11B.
model: "plugsky-vision-fast"plugsky-llama4
Paid
meta/llama-4-maverick-17b-128e-instruct
Context: 131K tokens
ToolsJSONLong-ctxMoE
Meta Llama 4 Maverick 17B (128 experts MoE, 128K ctx).
model: "plugsky-llama4"plugsky-qwen-vl
Paid
qwen/qwen3.5-397b-a17b
Context: 262K tokens
VisionMultimodalToolsJSONLong-ctxMoEReasoning
Qwen 3.5 397B MoE — multimodal + 256K context + reasoning.
model: "plugsky-qwen-vl"plugsky-kimi
Paid
moonshotai/kimi-k2.6
Context: 131K tokens
ToolsJSONLong-ctx
Long-context (256K) — MoonshotAI Kimi K2.6.
model: "plugsky-kimi"plugsky-longctx
Paid
mistralai/mistral-large-3-675b-instruct-2512
Context: 131K tokens
ToolsJSONLong-ctx
Mistral Large 3 675B — European, 128K context.
model: "plugsky-longctx"plugsky-mistral-medium
Paid
mistralai/mistral-medium-3.5-128b
Context: 131K tokens
ToolsJSONLong-ctx
Mistral Medium 3.5 128B — fast 128K context.
model: "plugsky-mistral-medium"plugsky-mistral-small
Paid
mistralai/mistral-small-4-119b-2603
Context: 131K tokens
ToolsJSONLong-ctx
Mistral Small 4 119B — fast + 128K context.
model: "plugsky-mistral-small"plugsky-nano
Paid
nvidia/nemotron-3-nano-30b-a3b
Context: 1M tokens
ToolsJSONLong-ctxMoE
NVIDIA Nemotron 3 Nano 30B MoE — 1M context, fast.
model: "plugsky-nano"🌸 Paid tier — Google Gemma
Compact multimodal models from Google. Lightweight, fast, with image input support.
plugsky-gemma-4
Paid
google/gemma-3n-e4b-it
Context: 32K tokens
MultimodalJSON
Google Gemma 3 Nano 4B — fast + multimodal.
model: "plugsky-gemma-4"plugsky-gemma3-nano-2b
Paid
google/gemma-3n-e2b-it
Context: 32K tokens
MultimodalJSON
Google Gemma 3 Nano 2B — ultra-compact, fast.
model: "plugsky-gemma3-nano-2b"plugsky-gemma3-nano-4b
Paid
google/gemma-3n-e4b-it
Context: 32K tokens
MultimodalJSON
Google Gemma 3 Nano 4B — small + fast + multimodal.
model: "plugsky-gemma3-nano-4b"🔢 Embeddings
Vector embeddings for semantic search, RAG, clustering, recommendations. Each model outputs a different dimension — check before switching.
plugsky-embed
Embed
nvidia/nv-embed-v1
Dim: 4096 · Ctx: 8K
Embeddings
NVIDIA NV-Embed v1 — best general embeddings.
model: "plugsky-embed"plugsky-embed-nim
Embed
nvidia/nv-embed-v1
Dim: 4096 · Ctx: 8K
Embeddings
NVIDIA NV-Embed v1 — best general embeddings (alias).
model: "plugsky-embed-nim"plugsky-embed-multilingual
Embed
baai/bge-m3
Dim: 1024 · Ctx: 8K
EmbeddingsMultilingual
BGE-M3 — multilingual embeddings (100+ languages).
model: "plugsky-embed-multilingual"⚡ API usage
OpenAI-compatible. Any model above can be used with a single line change. No code rewrites between models.
cURL
curl https://api.plugsky.com/v1/chat/completions \
-H "Authorization: Bearer sk-live-..." \
-H "Content-Type: application/json" \
-d '{
"model": "plugsky-pro",
"messages": [{"role": "user", "content": "Hello!"}],
"max_tokens": 200
}'
Python (openai SDK)
from openai import OpenAI
client = OpenAI(
api_key="sk-live-...",
base_url="https://api.plugsky.com/v1"
)
resp = client.chat.completions.create(
model="plugsky-pro",
messages=[{"role": "user", "content": "Hello!"}],
max_tokens=200,
)
print(resp.choices[0].message.content)
JavaScript / TypeScript
const r = await fetch("https://api.plugsky.com/v1/chat/completions", {
method: "POST",
headers: {
"Authorization": "Bearer sk-live-...",
"Content-Type": "application/json"
},
body: JSON.stringify({
model: "plugsky-pro",
messages: [{ role: "user", content: "Hello!" }],
max_tokens: 200
})
});
const data = await r.json();
console.log(data.choices[0].message.content);