Dify with unoblox: India AI ops in rupees

Dify + unoblox: open-source AI ops at ₹-native rates

Dify's model provider settings accept any OpenAI-compatible endpoint. Point it at unoblox, pick a model — Qwen, DeepSeek, or Claude — and run production workflows billed monthly in rupees on a GST invoice from an Indian entity, with input tax credit claimable.

Connect in 3 clicks

1. Sign in at unoblox.ai/sign-in and copy your API key ub-gw-…
2. In Dify, go to Settings → Model providers → OpenAI-compatible:
   • Provider name: unoblox
   • Base URL: https://api.unoblox.ai/v1
   • API Key: paste ub-gw-…
   • Model: pick from the catalog, e.g. qwen/qwen3-235b-a22b-instruct-2507 or deepseek-ai/deepseek-v3.2
3. Save, create a workflow, select unoblox as the LLM provider, and deploy.

Why Dify teams pick unoblox over a direct model account

	unoblox + Dify	A direct model account
Billing	One monthly ₹ GST invoice	Card billing, no Indian GST invoice
Model catalog	40+ models behind one key	Usually one vendor's lineup
Switching models	Change one field in Dify	New account, new key, new billing
Self-hosted Dify	Fully supported	Fully supported

Best Dify workflows on unoblox

• Content review: intake from Slack or email → Qwen summarization node → post to Discord or Airtable
• Support automation: customer query → DeepSeek classification → ticket assignment by category
• Data pipeline: CSV upload → Qwen extraction and entity tagging → database insert, one invoice for the whole run
• Knowledge-base Q&A: Dify's retrieval node + free Qwen3-Embedding-0.6B for vectors + Claude Sonnet for the final answer

Pricing examples (₹ per 1M tokens, input / output)

• Qwen3.8-27B ₹16.32 / ₹48.96 — vision, general-purpose, a good default
• DeepSeek V3.2 ₹26.21 / ₹38.30 — reasoning-heavy workflows
• Qwen3 1.7B ₹0 / ₹0 — free tier, rate-limited, good for prototyping a workflow before scaling it

Frequently asked questions

Can I use Dify's knowledge base with unoblox? Yes. Dify's retrieval pipeline works unchanged — embeddings and the answering LLM can sit on different unoblox models in the same workflow.

Do I need to rebuild workflows if I switch models? No. Change the model field in the provider settings; every workflow using that provider switches with it. No node-by-node edits needed.

Can I run self-hosted Dify with unoblox? Yes. Self-hosted Dify points at the unoblox gateway exactly like Dify Cloud does — same base URL, same key.

Does Dify support streaming with unoblox models? Yes. All unoblox chat models support streaming; Dify's UI shows tokens as they arrive.

Can I set a custom system prompt per workflow? Yes. Set it in Dify's LLM node's system-prompt field — it overrides any provider default for that workflow only.

How do I track unoblox spend against a specific Dify workflow? Dify logs token counts per node. Cross-reference the timestamps against your unoblox dashboard, which breaks down cost in ₹ by model and request.

Get started in rupees → https://unoblox.ai/sign-in