Models

The model catalogue, with IDs, context windows, and per-token pricing in ₹ INR.

Pass the model ID (the slug shown below) in the model field of your request. Prices are per 1,000 tokens for text and embeddings, and per image or per minute for image and speech models. Free-plan accounts can call the models marked Free; Developer and Pro plans unlock the rest.

{
  "model": "gpt-oss-20b",
  "messages": [{ "role": "user", "content": "Summarise this in one line." }]
}

Text Generation

Model ID	Context	Input / 1K Tokens	Output / 1K Tokens	Plan
`gpt-oss-20b`	128K	₹0.01	₹0.03	Free
`mistral-small-24b`	128K	₹0.02	₹0.04	Developer
`qwen3-6-27b`	256K	₹0.06	₹0.39	Developer
`llama-3-70b-instruct`	128K	₹0.10	₹0.10	Developer
`gpt-oss-120b`	128K	₹0.02	₹0.06	Pro
`qwen2-5-vl-72b`	32K	₹0.13	₹0.13	Pro
`qwen3-5-397b`	256K	₹0.09	₹0.52	Pro
`gemma-4-26b-a4b`	256K	₹0.02	₹0.04	Developer
`paddleocr-vl`	16K	₹0.02	₹0.02	Developer
`glm-5-2`	1M	₹0.19	₹0.58	Developer
`deepseek-v4-1-flash`	1M	₹0.04	₹0.16	Developer
`deepseek-v4-pro`	1M	₹0.18	₹0.52	Developer
`kimi-k2-7-code`	256K	₹0.13	₹0.52	Developer
`kimi-k3`	1M	₹0.41	₹1.96	Pro
`minimax-m3`	512K	₹0.08	₹0.31	Developer

Embeddings, Image & Audio

Model ID	Type	Price	Plan
`bge-multilingual-gemma2`	Embeddings	₹0.01/1K tokens	Developer
`qwen3-embedding-4b`	Embeddings	₹0.02/1K tokens	Developer
`stable-diffusion-xl`	Image	No Charge	Developer
`whisper-large-v3`	Speech-to-text	₹0.35/min	Developer
`whisper-large-v3-turbo`	Speech-to-text	₹0.35/min	Developer
`kokoro-tts`	Text-to-speech	₹0.15/1K chars	Developer

“Free” models are available on every plan. “Developer” models require a Developer or Pro account; “Pro” models require a Pro account. See Pricing & Billing for how charges are calculated, or the pricing page for plan details.

Listing Models from the API

GET https://api.nabh.cloud/v1/models returns every model the API serves, in the OpenAI list format (client.models.list()), and GET https://api.nabh.cloud/v1/models/<id> returns one (client.models.retrieve("gpt-oss-20b")). Both need an API key. A retired ID returns the model that replaced it; an unknown ID returns 404 with the code INVALID_REQUEST.