Awesome Free LLM APIs IR

Choosing a Free LLM API in Iran: From Signup to First Request

Practical guide for selecting, signing up, testing connectivity, and making your first free LLM API request from inside Iran. Covers security, quota types, and access evidence.

Last reviewed: 2026-07-18

Quick answer: Pick a provider whose free-tier type is clearly documented, does not require a credit card for account creation, has a timestamped Iran-access status, and ideally supports the OpenAI-compatible API standard. No single provider is best for every project.

The up-to-date list of providers, quotas, and Iran-access status is in the Free LLM API Catalog. Always check the last-review date and the provider's dedicated page before committing — service policies can change.

Why the selection process is different for Iranian users

In a typical comparison, model quality, speed, and price matter most. For users inside Iran, several additional hurdles exist:

  • The provider's website may load, but the signup form or SMS verification may not work.
  • Account creation may succeed, but model inference may be rejected due to regional restrictions.
  • Some services officially do not support Iran, even if network connectivity is temporarily available.
  • "Free" could mean permanent quota, free models, monthly credit, or a short trial.
  • Rate limits, token caps, and concurrency on free accounts are usually lower.

This project separates statuses into "verified working (direct)", "verified working (VPN)", "signup blocked", "verified blocked", and "unknown".

Step 1: Define your use case precisely

Before comparing providers, clarify what you need the model for.

Chat and Persian text generation

For chatbots, summarization, and content generation:

  • Persian response quality and instruction following
  • Context window length
  • Time to first token
  • Output length limits
  • Data retention policy

Coding and agents

For coding assistants or agents, verify these capabilities beyond code quality:

  • Tool calling
  • Structured output or JSON schema
  • Endpoint stability
  • Concurrent request limits
  • OpenAI SDK compatibility

RAG and semantic search

For RAG, confirm the provider actually offers a dedicated Embedding endpoint. Also check vector dimensions, token limits, Persian language support, batch size, and data policy.

Student projects and prototypes

For prototypes, prioritize:

  1. Simple signup
  2. No credit card required
  3. Predictable quota
  4. Independent API key
  5. Clear documentation
  6. Easy provider switching

Step 2: Understand what "free" really means

"Free API" does not mean the same thing everywhere.

Permanent quota

A portion of usage is free under current policy, usually reset on a schedule. "Permanent" does not guarantee the current cap forever — providers can change terms.

Free models

Only specific models may be free. Newer or more capable models may cost extra or consume different quotas.

Free credit

The account receives a limited credit balance. Once exhausted, continued usage may require payment.

Trial

Trials are suitable for short-term evaluation but are not a reliable foundation for a long-running project.

See the detailed guide: Free Tier vs Trial vs Credit

Step 3: Interpret Iran-access status correctly

A provider's homepage loading does not mean the API is usable from Iran. A valid test should cover:

  1. Access to the signup page
  2. Account creation or login
  3. API key generation
  4. Listing authorized models
  5. Sending a real inference request
  6. Receiving a valid response

A public endpoint ping, DNS resolution, or model list page alone is insufficient.

"Unknown" does not mean the service works — it means there is not enough recent evidence for a definitive conclusion.

Step 4: Check OpenAI compatibility

OpenAI-compatible means the provider uses a request structure similar to OpenAI's API. You can usually use the same SDK with a changed base_url and model. It does not mean the provider uses OpenAI infrastructure or supports every feature identically.

Differences may exist in:

  • Model names
  • Tool calling
  • Streaming
  • Structured output
  • Embedding
  • Generation parameters
  • Error formats

Related list: OpenAI-Compatible APIs Without a Credit Card

Step 5: Sign up securely

Create a separate API key for each project to simplify key revocation and usage monitoring.

Essential tips:

  • Never hardcode keys in source code.
  • Do not commit .env files.
  • Do not post API keys, cookies, recovery codes, or account dashboard screenshots in public issues.
  • Use CI secrets for public repositories.
  • Rotate keys immediately if a leak is suspected.

Sample .gitignore:

.env
.env.*
!.env.example
.venv/
node_modules/

Step 6: Make your first request

If the provider is OpenAI-compatible, read settings from environment variables to simplify switching.

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["LLM_API_KEY"],
    base_url=os.environ["LLM_BASE_URL"],
    timeout=30.0,
    max_retries=2,
)

response = client.chat.completions.create(
    model=os.environ["LLM_MODEL"],
    messages=[
        {"role": "system", "content": "Answer briefly and precisely."},
        {
            "role": "user",
            "content": "Name three applications of language models in education.",
        },
    ],
)

print(response.choices[0].message.content)

Set environment variables:

export LLM_API_KEY="..."
export LLM_BASE_URL="https://provider.example/v1"
export LLM_MODEL="MODEL_ID"
python app.py

Get the Base URL and Model ID from the provider page and official docs; do not copy sample values without verification.

Step 7: Evaluate the result

A successful test means more than HTTP 200. Record:

  • Time and network type
  • Signup success
  • Key generation success
  • Model ID used
  • Real request result
  • Approximate latency
  • Sanitized error
  • Test date

Do not publish sensitive account information or private prompt text.

Final checklist

Before depending on a provider, answer these questions:

  • Does the free tier align with your project's lifespan?
  • Is the required model available on the free account?
  • Are RPM, RPD, or TPM limits sufficient for your load?
  • Is the Iran-access status recent and documented?
  • Are signup and real requests tested?
  • Are timeout and retry configured?
  • Do you have a fallback provider?
  • Are secrets stored outside the repository?

When to switch providers

Consider switching when:

  • Regional or signup policy has changed.
  • Quota is insufficient for actual usage.
  • 429 or 5xx errors are persistent.
  • Required capabilities (e.g., tool calling) are unreliable.
  • Persian response quality does not meet your needs.
  • Data and privacy terms are incompatible with your project.

Using a common interface with LLM_BASE_URL, LLM_API_KEY, and LLM_MODEL environment variables reduces migration cost.

Summary

For an Iranian user, the best API is not necessarily the most famous model. The right choice is a service with clear terms, a verifiable free tier, tested network access, and easy replaceability when policies change.

For the current provider comparison, use the live catalog.

If you have tested a service from Iran, submit the result in an Iran Access Report issue. Do not include any keys, cookies, or personal information in the report.

If this resource is useful, star the repository on GitHub to help others find it.

Compare all APIsEnglish Guides