Free LLM API Tokens

SambaNova Cloud free plan — published caps and card notes

SambaNova Cloud is an official, OpenAI-compatible API. Our row records a per-model daily token cap and notes that the free plan has required a card and a purchase since mid-October.

Last updated 12 Nov 2026. The provider figures on this page follow the tracker snapshot of 12 Nov 2026, 08:59 IST.

What is SambaNova Cloud on this tracker?

SambaNova Cloud is listed as an official provider with OpenAI compatibility Yes. The recorded free allowance is “200,000 tokens/day per model (as of 3 Oct)”. Tokens per day are 200,000. Limits are “20 requests/day”. Card required is Yes. Relay warning: no. Our notes say: Since 15 Oct the Free plan requires a card and a purchase.

Those notes matter as much as the daily token figure. The official URL we store is https://cloud.sambanova.ai/plans. The entry was first listed on 3 Oct 2026 and last updated on 15 Oct 2026 — later than many neighbouring official rows, because of the card-and-purchase change recorded in the notes. List snapshot: 12 Nov 2026, 08:59 IST.

Snapshot from this trackerFigures below are copied from our provider list (entry last updated 15 Oct 2026). They change often — confirm on the official site before you rely on them.
Official siteVisit SambaNova Cloud
CategoryOfficial provider
OpenAI compatibleYes
Free allowance200,000 tokens/day per model (as of 3 Oct)
Tokens per day200,000
Rate limits20 requests/day
Card requiredYes
Base URLhttps://api.sambanova.ai/v1
Relay warningNo
First listed3 Oct 2026
Entry updated15 Oct 2026
NotesSince 15 Oct the Free plan requires a card and a purchase.

How should I read “per model” together with a request cap?

The daily token figure we store is described as applying per model, as of the official wording we saw on 3 October. That is not necessarily a single shared pool, and it is not a licence to assume every current model is in the free plan. Official plans may add, remove or re-price models later.

Separately, we record a limit of 20 requests/day. A 20-requests-per-day cap, if still accurate on the official page, can stop a coding assistant long before the token cap does. Always compare both fields, and confirm both on the official plans page. Background: Rate limits explained.

What changed in the card requirement?

Our notes record that since 15 October the Free plan requires a card and a purchase. The card field is therefore Yes. We do not invent the size of that purchase, or whether a past no-card account was grandfathered; those details, if they exist, belong on the official site. If you signed up earlier, do not assume our older memory of a no-card plan still applies — look at the live plans page.

This is an example of why the tracker stores an entry-level last_updated date and a changelog on the home page, instead of treating every official free tier as frozen.

How do I call the published base URL?

The base URL we record is https://api.sambanova.ai/v1. Use a SambaNova key in the environment, the official OpenAI SDK, and a model id from SambaNova’s official documentation.

Python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["API_KEY"],
    base_url="https://api.sambanova.ai/v1",
)

resp = client.chat.completions.create(
    model="MODEL_ID",  # copy the model id from the provider's docs
    messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)

JavaScript

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.API_KEY,
  baseURL: "https://api.sambanova.ai/v1",
});

const resp = await client.chat.completions.create({
  model: "MODEL_ID", // copy the model id from the provider's docs
  messages: [{ role: "user", content: "Hello" }],
});
console.log(resp.choices[0].message.content);

curl

curl https://api.sambanova.ai/v1/chat/completions \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"MODEL_ID","messages":[{"role":"user","content":"Hello"}]}'

The MODEL_ID string is not guessed here. Copy it from the provider’s dashboard or official model list, because catalogues change.

If the official console requires you to attach a card or complete a purchase before the key can call a model, the SDK will not bypass that. Handle 401, 403 and payment errors by reading the official account page, not by retrying in a loop.

Is SambaNova a relay?

No. It is official in our category, with no relay warning. You still accept SambaNova’s terms when you create an account. For private prompts, official providers are the default we recommend over flagged relays, as explained in Official providers vs gateways.

How does this compare with Groq and Cerebras?

In our file, Groq also has a recorded daily token figure of 200,000 on listed models, with a much higher recorded daily request cap and a card field of No — see Groq and confirm both official pages, because those numbers can change. Cerebras records a higher daily token figure but only during a 30-day card-required trial — see Cerebras. SambaNova’s distinctive recorded constraints are the per-model wording, the 20 requests/day limit, and the card and purchase note on the free plan.

None of these comparisons is a quality ranking. They are a reading of published fields.

How do I verify this row?

Open https://cloud.sambanova.ai/plans and check the allowance, the per-model language, the request cap, the card rule and the base URL. If something has changed, send the official link. We will not drop the card note to make the row look easier, and we will not convert “per model” into a shared pool unless an official page says that.

What should I do on signup day?

Open the official plans page first, not only this guide. Confirm whether a card and a purchase are still required, which models are in the free plan, and whether the 20-requests-per-day figure still holds. Then create the account, store the key in the environment, and make one short request with an official model id.

If you are doing this to run an editor assistant, do the maths in requests, not only in tokens: twenty official requests is a small number for an agent that retries. You may want a provider with a higher published request cap for that workflow, or you may want to use SambaNova only for single, careful calls.

How should teams record this provider internally?

Write down the official plan name, the date you confirmed the card-and-purchase rule, and the two caps (tokens per model per day, requests per day) as they appeared on the official page. Link to the official plans URL. You can also link to this tracker row, but treat our snapshot date as a hint to re-check, not as a contract. If SambaNova’s official wording changes, update your internal note even if this page has not yet caught up.