EN ▾
Uncensored CodingCode without refusals.Get API key
per 1M input tokens
$0.15
Output tokens / 1M
$0.50
token context
100,000
trial credit
$0.50
requests per minute
300
uncensoredcodingai.com

Uncensored Coding AI Free: Trial, Costs and Options

There is no unlimited free tier here, and we say so up front. You can try the API with a $0.50 trial, then pay per token. This page works out what that buys for real coding work.

Updated

Terminal with flowing code beside a small stack of prepaid tokens, showing a free trial then pay-per-token coding

Key points

  1. The free part is a one-click $0.50 trial, one per person, valid for 7 days.
  2. After that you pay $0.15 per 1M input tokens and $0.50 per 1M output tokens from prepaid credit.
  3. A typical code-review call costs under a cent, so $0.50 covers hundreds of calls.
  4. Top-ups start at $20 and are paid by crypto (USDT on TRC20, USDC on Base).

Is it free? The honest answer

People who search for an uncensored coding ai free option usually want one of two things: a place to try an unrestricted model before paying, or a permanent free tier. We can offer the first and not the second. New users can claim $0.50 of trial credit with one click on the API key page after sign-up. It is one trial per person, valid for 7 days, and it needs no payment details. When it runs out, the service continues on prepaid credit, and you pay only for the tokens you use.

Why not a free unlimited plan? Running a large language model costs real money in GPU time for every request, and an unlimited free tier attracts automated abuse that crowds out genuine users. Charging per token keeps the price low and the service predictable. The rates are $0.15 per 1M input tokens and $0.50 per 1M output tokens. There is no subscription, failed or refused requests are not charged, and credit never expires.

Below you will find what $0.50 actually buys in coding work, how to hook the API into the tools developers use, which free routes exist if you need zero spend, and where our model is the wrong pick.

What $0.50 buys in coding work

Coding prompts are input-heavy: you paste files, diffs and stack traces, and the model replies with something shorter. Input tokens cost less than output tokens on this API, which suits that pattern. Here are three typical calls, priced exactly.

TaskTokens in / outCost per callCalls per $0.50
Explain a stack trace1,500 / 400$0.000425about 1,176
Review a diff or write a function with context6,000 / 1,500$0.00165about 303
One step of an agent loop with a large repo map30,000 / 2,000$0.0055about 90

The arithmetic for the middle row: input 6,000 x $0.15 / 1M = $0.0009, output 1,500 x $0.50 / 1M = $0.00075, total $0.00165. Divide $0.50 by that and you get roughly 303 calls. The first row: $0.000225 for input plus $0.0002 for output gives $0.000425. The last row: $0.0045 plus $0.001 gives $0.0055.

So the trial is generous for a hands-on evaluation. You can run a few dozen real tasks from your own codebase and judge the answers, which is far more informative than any demo. Keep in mind the trial key is limited to 2 concurrent requests and 60 requests per minute, which is fine for manual testing but will throttle an agent that fans out many tool calls at once.

A first call to test

Create a key on the API key page, claim the trial, and send a request. The endpoint follows the OpenAI chat format, so any OpenAI SDK works once you change the base URL and key. The model ID is uncensored.

curl https://api.uncensoredcodingai.com/v1/chat/completions \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "uncensored",
    "messages": [{"role": "user", "content": "Write a Python function that parses a CIDR range and lists its hosts."}],
    "max_tokens": 400
  }'

In Python, here is a small reviewer script that sends a diff and prints the estimated cost of the call from the returned usage numbers. Keeping that cost line in your own tools means you always know what a task cost.

from openai import OpenAI

client = OpenAI(
    base_url="https://api.uncensoredcodingai.com/v1",
    api_key="YOUR_KEY",
)

diff = open("change.diff").read()

r = client.chat.completions.create(
    model="uncensored",
    messages=[
        {"role": "system", "content": "You are a blunt code reviewer. List bugs first, style last."},
        {"role": "user", "content": "Review this diff:\n\n" + diff},
    ],
    temperature=0.2,
    max_tokens=1200,
)
print(r.choices[0].message.content)
u = r.usage
print("cost: $%.5f" % (u.prompt_tokens * 0.15 / 1e6 + u.completion_tokens * 0.50 / 1e6))

Streaming, tool calling and JSON mode are supported, and temperature, top_p, stop, seed and the penalties are passed through. Details are in the documentation. For code tasks a low temperature, around 0.2, tends to give steadier output.

Use it in the tools you already have

You do not need to write a client to use this. Several popular coding tools let you add a custom OpenAI-compatible endpoint. Aider accepts one through environment variables, and editor extensions such as Continue and Cline expose a base URL and key field in their provider settings. The pattern is the same everywhere: base URL, key, model name.

export OPENAI_API_BASE="https://api.uncensoredcodingai.com/v1"
export OPENAI_API_KEY="YOUR_KEY"

# the openai/ prefix tells aider to use the OpenAI-compatible route
aider --model openai/uncensored

Three practical notes when you wire up an agent-style tool:

  • Watch the context. The window is 100,000 tokens for prompt plus completion. Tools that stuff in a whole repository can fill it quickly, and larger prompts cost proportionally more.
  • Mind concurrency. A full key allows 8 concurrent requests and 300 per minute. Over a limit you get HTTP 429 with the code rate_limited or concurrency. Configure the tool to retry, or lower its parallelism.
  • Keep the key in an environment variable, not in a committed config file.

There is also a longer overview of the idea in our article on an uncensored LLM API for coding.

What the paid route costs per month

Once the trial ends, the question becomes how much a steady habit costs. Use the numbers above and your own volume.

  • Light use: 100 diff reviews per day at $0.00165 each is $0.165 a day, or $4.95 over 30 days.
  • Heavy agent use: 400 agent steps per day at $0.0055 each is $2.20 a day, or $66 over 30 days.

Top-ups are any whole amount from $20 to $500, with a bonus of +5% from $50, +10% from $100 and +15% from $200. A $100 top-up becomes $110 of credit. At the light-use rate, a plain $20 top-up covers about 121 days, and because credit never expires you are not racing a billing cycle.

Payment is by crypto (USDT on TRC20, USDC on Base). If you do not hold either, you will need to obtain one through an exchange before topping up, so allow time for that on the first purchase. The pricing page lists the current options and bonus tiers.

Free routes if you need to spend nothing

If your budget is zero, there are legitimate free routes, and it is better to know them than to be disappointed here.

  • Run an open-weight model locally. Tools that serve open-weight models on your own machine are free to use once you have the hardware. Quality and speed depend on your GPU and memory, and larger models need serious hardware.
  • Use the free tiers of mainstream assistants for ordinary coding questions. Their limits and rules change, so check each provider's current terms.
  • Use the trial here once, on your hardest real tasks, to see whether an unrestricted hosted model is worth paying for.

Be careful with sites that advertise free unlimited access to an unrestricted model through shared keys or unofficial proxies. Prompts that contain your source code may be logged, resold or leaked. On our service, prompts are not used for training.

When this is the right tool

The model is open-weight, run by an independent provider, and tuned to answer without content refusals for lawful adult use. For code, that matters in a few specific places: security research on systems you are authorised to test, writing parsers and scrapers that a cautious model hesitates over, red-team tooling for your own infrastructure, and generating test data that a stricter model refuses to produce. It will not help with anything unlawful, and one rule is absolute: no sexual content involving minors.

It is also fair to say where it is not the best fit. It is text only, with no image or audio generation, embeddings or fine-tuning. It is not the official API of any big lab, so if your work needs a specific vendor's model, use that vendor. And benchmarks for coding quality vary by task, so run your own tests with the trial before moving production work. A focused evaluation of 20 or 30 tasks you really do costs a few cents and tells you more than a leaderboard.

A simple way to run that evaluation: collect ten tasks from your last month of work, such as a flaky test, a regex you hated writing, a migration script and a small refactor. Run each one with the same prompt you would give a colleague, save the answers, and mark each as usable as is, usable after edits, or wrong. If most land in the first two groups, the model earns a place in your toolbox for that kind of task. If not, you have spent a few cents to find out, and the unused trial credit simply expires after its 7 days.

Keep temperature low for code and set max_tokens so a runaway answer cannot drain credit. If you need machine-readable output, JSON mode with response_format set to json_object keeps replies parseable, which helps when a script consumes the result.

Questions and answers

Is there a free uncensored coding model here?

There is a free trial, not a free tier. You can claim $0.50 of credit with one click after sign-up, valid for 7 days, then pay per token.

How many requests does the trial cover?

At 6,000 tokens in and 1,500 out, a call costs $0.00165, so $0.50 covers about 303 calls. Larger agent steps cost more per call and cover fewer.

Can I pay by bank transfer?

Top-ups are by crypto (USDT on TRC20, USDC on Base), in any whole amount from $20 to $500.

Does the trial need payment details?

No. You sign up with Google or with an email and password, then claim the trial on the API key page. One trial per person.

Your key is one form away

Create an account, copy the key, change the base URL. That is the whole setup.

Get API key