Create an account and top up
Register, then add credit. The minimum top-up is $10 and the balance does not expire while the account is active.
Free AI models on TokenRa right now
A model counts as free here when the live pricing API reports $0 for both input and output tokens. On October 31, 2026 that is nemotron-3.5-super-vl-free.
| Model | Model ID | Input / output per 1M | Accepts | Context | Free until |
|---|---|---|---|---|---|
| Nemotron 3.5 Super VL | nemotron-3.5-super-vl-free | $0 / $0 | Text, image | 1M tokens | No end date announced |
Source: GET https://tokenra.io/api/pricing, where a token-priced model with model_ratio 0 costs nothing per token. The context window and the accepted input types were checked with live requests on October 31, 2026.
What “free” means: $0 per token, funded account
Free models cost $0 per token, but TokenRa has no free signup credit and no separate free tier, so a free model is not a free trial.
You need a funded account first
A new account starts at $0. On October 31, 2026 we sent a request for nemotron-3.5-super-vl-free from a new account with a $0 balance and got HTTP 403 with insufficient_user_quota, the same answer a paid model gives. The minimum top-up is $10 and credits do not expire while the account is active, so the first top-up also covers every paid model on TokenRa.
- What $0 covers: input, output and cached input tokens, with no subscription and no platform fee.
- Promotional: a free offer can end or change without notice, and the live console price is authoritative. Account rate limits still apply.
- Same key and endpoint as paid models: a free model is called exactly like a paid one, so switching to a paid fallback means changing only the model ID.
Check which models are free right now (Python 3)
curl -s https://tokenra.io/api/pricing | python3 -c "
import sys, json
for m in json.load(sys.stdin)['data']:
if m['quota_type'] == 0 and m['model_ratio'] == 0:
print(m['model_name'])"How to use a free model through the API
A free model uses the same OpenAI-compatible endpoint as every other model. The two snippets below were run as written on October 31, 2026 and returned HTTP 200.
Create an API key
Open the dashboard and generate a key. Keep it on the server, not in browser code.
Send a request
Use the model ID nemotron-3.5-super-vl-free on the OpenAI-compatible endpoint.
cURL
curl https://tokenra.io/v1/chat/completions \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"nemotron-3.5-super-vl-free","messages":[{"role":"user","content":"Name three uses of a free AI model API, one short line each."}]}'Python (OpenAI SDK)
from openai import OpenAI
client = OpenAI(
base_url="https://tokenra.io/v1",
api_key="YOUR_API_KEY",
)
resp = client.chat.completions.create(
model="nemotron-3.5-super-vl-free",
messages=[{"role": "user", "content": "Give me three project ideas I can build with a free AI model API."}],
)
print(resp.choices[0].message.content)The same key works for every other model. The API quickstart covers the base URL, authentication, errors and retries for every model.
Cheapest AI models when nothing is free
Free offers end, and a funded account is needed either way, so these are the lowest-priced chat models on TokenRa today. At the lowest row, $0.05 per 1M input tokens, a $10 top-up covers about 200 million input tokens, or about 51 million output tokens at $0.194 per 1M.
| Model | Model ID | Input | Output | 10M input + 2M output |
|---|---|---|---|---|
| GLM-5.3 Flash | glm‑5.3‑flash‑discounted | $0.05 | $0.194 | $0.89 |
| DeepSeek V4.1 Flash 1017 | deepseek-v4.1-flash-1017-0.2x | $0.06 | $0.24 | $1.08 |
| Qwen3.8 Flash | qwen3.8-flash | $0.08 | $0.235 | $1.27 |
| DeepSeek V4.1 Flash | deepseek-v4.1-flash-50off | $0.09 | $0.36 | $1.62 |
| MiniMax M3.1 Flash | minimax-m3.1-flash | $0.09 | $0.36 | $1.62 |
| Omen Alpha | omen-alpha | $0.13 | $0.50 | $2.30 |
Some models have several IDs at different prices, for example a discounted ID next to the standard one; each row shows the cheapest ID of its model. The model name links to its API docs, which explain the variants. glm‑5.3‑flash‑discounted is written with non-breaking hyphens (U+2011), so copy the ID from this page. Prices change; the live console is authoritative.
Free AI models FAQ
Which AI models are free on TokenRa?
As of October 31, 2026, nemotron-3.5-super-vl-free is priced at $0 per 1M input and output tokens. No end date has been announced for the $0 rate. Every other model on TokenRa is billed per token.
Is Nemotron 3.5 Super VL really free?
On TokenRa, Nemotron 3.5 Super VL costs $0 per 1M input, output and cached input tokens, with no subscription. The $0 rate applies to your usage, but your account still needs a positive balance to send requests.
Do I need a credit card or a top-up to use a free AI model?
You need a funded account. TokenRa gives no free signup credit, and on October 31, 2026 a request from a new account with a $0 balance was rejected with HTTP 403 and the error code insufficient_user_quota, even for a $0 model. The minimum top-up is $10 and credits do not expire while the account is active. After that, calls to a $0 model do not draw the balance down.
Does TokenRa give free credits to new accounts?
No. A new account starts with a $0 balance and there is no trial allowance or separate free tier. The $0 models above are the free option, and every other model is billed per token.
What happens when the free period ends?
No end date has been announced for the $0 rate. When a free offer ends, TokenRa bills the model at the rate shown in the console, so check the live rate before you rely on it and keep a fallback from the cheapest-models table, because free offers are promotional and can change without notice.
How can I check which models are free right now?
Request https://tokenra.io/api/pricing and list the models whose quota_type is 0 and whose model_ratio is 0. Those are billed at $0 per token. The snippet in the section above does exactly that.
Can I use a free model in production?
You can, but plan for the offer to end. Free access is promotional and can end or change without notice, and account rate limits still apply. Route production traffic through a paid fallback model from the cheapest-models table so that a price change does not become an outage.
Is the free model OpenAI compatible?
Yes. Send requests to https://tokenra.io/v1/chat/completions with your TokenRa key and the model ID nemotron-3.5-super-vl-free. Existing OpenAI SDK clients work after you change base_url and model.
What are the cheapest AI models on TokenRa?
As of October 31, 2026, the lowest input prices per 1M tokens are GLM-5.3 Flash ($0.05), DeepSeek V4.1 Flash 1017 ($0.06), Qwen3.8 Flash ($0.08). The table above lists output prices and the cost of an example workload of 10M input and 2M output tokens.
Is TokenRa the developer of these models?
No. The models are developed by their vendors. TokenRa is an independent API gateway that routes requests to enabled models and is not the model developer or owner.
Prices are TokenRa console rates from the public pricing API on October 31, 2026. The model names and any related marks belong to their rights holders. TokenRa is an independent API gateway and this page is not an official vendor page or an endorsement.