Free LLM API — no card required

Free LLM API, no payment wall

Most "free tiers" quietly demand a card on file before you can call the API — the free part only kicks in after that gate. unoblox's free tier (Qwen3 1.7B) skips the gate entirely: no card, no international payment method, no expiring trial. You sign up with email and start calling the endpoint.

What "free" hides elsewhere, and what it means here

• Card-gated free tiers: you still need a payment method capable of international billing before the free credits activate.
• Time-boxed trials: free for 14 or 30 days, then a card is mandatory.
• Free-but-metered surprises: a generous quota that quietly starts billing once you cross it, sometimes without a clear warning.

unoblox's free tier avoids all three: ₹0, no card, no countdown — just a fair-use rate limit and a monthly usage cap so the tier isn't abused for resale.

The model behind it

Model: Qwen3 1.7B Instruct, Alibaba's small instruction-tuned model, optimized for low-latency serving rather than frontier reasoning. Endpoint: the same OpenAI-compatible /v1/chat/completions used by every other model on unoblox — no separate SDK. Pricing: ₹0 input, ₹0 output. Your monthly invoice shows the line item at zero. Rate limit: 100 requests/minute per key, which is comfortably above what most early-stage apps generate.

Quickstart (no card, three steps)

1. Sign up at https://unoblox.ai/sign-in with email and password.
2. Create an API key (ub-gw-…) from the dashboard.
3. Call it:

from openai import OpenAI

client = OpenAI(api_key="ub-gw-YOUR_KEY", base_url="https://api.unoblox.ai/v1")
resp = client.chat.completions.create(
    model="qwen/qwen3-1.7b",
    messages=[{"role": "user", "content": "Summarize this in two lines: ..."}]
)
print(resp.choices[0].message.content)

Your first monthly statement arrives showing ₹0 due — nothing to reconcile, nothing to cancel.

Free tier vs a paid model, what actually changes

Dimension	Free (Qwen3 1.7B)	Paid (e.g. Llama 4 Scout)
Cost per 1M tokens	₹0 / ₹0	₹10.08 / ₹30.24
Context window	32K tokens	128K tokens
Best for	Classification, short generation, prototyping	Higher-accuracy production workloads
Support	Community, docs	Email support
Rate limit	100 req/min	Configurable per key

Nothing else changes — same endpoint, same key format, same SDK. Switching later is a one-line model change.

When free genuinely stays free

Prototyping, internal tools, low-traffic classification or summarization endpoints, and hackathon-style projects rarely approach the rate limit or the monthly cap. If your usage is bursty rather than constant, the free tier can outlast the project itself.

When to move off it

Move to a paid model once you need: consistently higher accuracy on open-ended tasks, a longer context window than 32K, or predictable throughput above the shared rate limit. All three are one config change away — no re-integration.

Frequently asked questions

Why is this free if the model is genuinely useful? Small instruction-tuned models are cheap enough to run at scale that unoblox can offer them at ₹0 without subsidizing a loss — it's also how new users try the platform before committing spend to bigger models.

Is there a sunset date for the free tier? No planned one. unoblox reserves the right to introduce stricter abuse limits if the tier is misused, but the ₹0 price itself isn't time-limited.

Can I use it for a commercial product? Yes, within the rate limit and monthly cap. Many early-stage products run real traffic on it.

Do I need a business account or GSTIN to use the free tier? No. A personal email is enough to start; add GST details later if you upgrade and need ITC-eligible invoices.

What counts as abuse? Reselling API access, running synthetic content farms, or scripted scraping at volume. Normal application traffic, even fairly busy traffic, isn't a problem.

Can I migrate from another provider's free tier to this one? Yes — it's a two-line change: swap base_url to https://api.unoblox.ai/v1 and use your ub-gw-… key.

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