Aider with unoblox (India)

Aider is the terminal-based pair-programming tool that edits your repository directly from chat instructions. It talks to any OpenAI-compatible endpoint, so pointing it at unoblox gives you a choice of coding-capable models, including a code-specialised one, billed per token in rupees instead of a foreign subscription.

Why run Aider against unoblox

Aider was built around swapping models freely, since different models suit different edit sizes and languages. unoblox's single endpoint and key make that swap a command-line flag rather than a new account: the same repository session can start on a cheap model for small fixes and move to a stronger one for a gnarly refactor, with both calls landing on the same rupee-denominated invoice.

Setting up the connection

Aider reads its backend from environment variables or command-line flags. Set the base URL and key once in your shell profile:

export OPENAI_API_BASE=https://api.unoblox.ai/v1
export OPENAI_API_KEY=ub-gw-xxxxxxxxxxxxxxxx

aider --model openai/gpt-4.1

Every model id in the unoblox catalog is a valid value for --model — change it per session depending on the task.

Matching a model to the kind of change

Kind of change	Model	₹ per 1M tokens (input / output)
Quick fixes, boilerplate	DeepSeek V4 Flash	₹9.07 / ₹18.14
Code-focused editing	Kimi K2.7	₹68.5 / ₹342.7
Open-weight option with vision	Qwen3.8-27B	₹16.32 / ₹48.96
Large, cross-file refactors	GPT-5	₹126 / ₹1008

For any other model in the catalog, including newer point releases, see live ₹ pricing on /models before pointing a long editing session at it.

Working with Aider's repo map and diffs sensibly

Aider sends your repository map and the relevant file contents as context on every request, so edit-heavy sessions on large files can use meaningfully more input tokens than a single chat question would. A practical pattern is to keep routine, well-scoped edits on a cheaper model, and reserve a stronger model for the moments where Aider's first attempt at a diff does not apply cleanly or the change spans several files at once.

Frequently asked questions

Does Aider need any special support for unoblox? No — Aider already supports any OpenAI-compatible endpoint through the standard base-URL and key environment variables shown above.

Can I switch models mid-session? Yes, Aider supports changing the active model inside a running session; the next request simply goes to the newly selected model id.

Will my source code leave India when I use an external model? Yes, for GPT, Claude, DeepSeek, and other external models, the file contents Aider sends as context are processed on that provider's own infrastructure; unoblox adds rupee billing and a GST invoice, not India data residency, for those models.

Is there a free model to test the Aider setup itself? Yes — Qwen3 1.7B is free (₹0), which is enough to confirm your environment variables and Aider configuration work before spending on a coding-capable model.

Does streaming work so I can see edits as they are generated? Yes, streaming responses work the same way they would against any OpenAI-compatible endpoint, and Aider uses it by default where supported.

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