Source: https://ailoot.ai/models/ms/nvidia-llama-3-1-nemotron-safety-guard-8b-v3/
Canonical: https://ailoot.ai/models/ms/nvidia-llama-3-1-nemotron-safety-guard-8b-v3/
Language: ms

# Llama 3.1 Nemotron Safety Guard 8B v3

[Markdown](https://ailoot.ai/models/ms/nvidia-llama-3-1-nemotron-safety-guard-8b-v3/index.md) · [JSON](https://ailoot.ai/models/ms/nvidia-llama-3-1-nemotron-safety-guard-8b-v3/facts.json)

## [Llama 3.1 Nemotron Safety Guard 8B v3](https://ailoot.ai/models/ms/nvidia-llama-3-1-nemotron-safety-guard-8b-v3/)

Nemotron Safety Guard v3 memeriksa mesej pengguna dan respons model berdasarkan taksonomi risiko yang ditetapkan. Mengembalikan JSON dengan penilaian keselamatan berasingan dan kategori pelanggaran.

- [**Ailoot · OpenAI Chat** `POST /v1/chat/completions` `nvidia/llama-3.1-nemotron-safety-guard-8b-v3`](https://ailoot.ai/models/ms/nvidia-llama-3-1-nemotron-safety-guard-8b-v3/#method-0)
- [**Ailoot · OpenAI Chat** `POST /v1/chat/completions` `nemotron-safety-guard`](https://ailoot.ai/models/ms/nvidia-llama-3-1-nemotron-safety-guard-8b-v3/#method-1)
- [**Ailoot · OpenAI Chat** `POST /v1/chat/completions` `nemotron-safety-guard-latest`](https://ailoot.ai/models/ms/nvidia-llama-3-1-nemotron-safety-guard-8b-v3/#method-2)

Tentang model

Huraian ini merangkumi keupayaan model. Input, output dan pilihan yang tersedia di Ailoot disenaraikan di bawah untuk kaedah yang dipilih.

#### Memeriksa kedua-dua pihak dalam dialog

Llama 3.1 Nemotron Safety Guard 8B v3 dibangunkan oleh NVIDIA untuk moderasi interaksi manusia dengan model bahasa. Mesej pengguna sahaja boleh diberikan atau ditambah dengan respons pembantu, termasuk beberapa ujaran. Model menentukan keselamatan setiap pihak: permintaan berbahaya dan penolakan pembantu yang betul mendapat penilaian berbeza. Ini membolehkan input diperiksa sebelum penjanaan dan respons yang telah disediakan dianalisis secara berasingan.

Hasilnya berupa JSON. Medan wajib `User Safety` mengandungi `safe` atau `unsafe`. Medan `Response Safety` muncul apabila input mengandungi respons pembantu. Jika pelanggaran dikesan, `Safety Categories` ditambahkan dengan senarai kategori yang dipisahkan oleh koma. Hasil sebegini sesuai untuk penapisan program dan penghantaran kes yang dipertikaikan untuk semakan tambahan.

#### Kategori dan konteks budaya

Arahan menetapkan peraturan dan nama kategori untuk semakan. Taksonomi yang diterbitkan mengandungi 23 kumpulan risiko: antaranya keganasan, ancaman, gangguan, pendedahan data peribadi, penipuan dan kod berniat jahat. Nama dalam hasil bergantung pada dasar yang diberikan, jadi nama itu perlu ditafsir bersama peraturan tugasan khusus.

Latihan mengambil kira ciri wilayah bagi topik sensitif. Data disesuaikan dengan ungkapan tempatan dan konteks budaya, manakala terjemahan yang tidak konsisten dibuang. Persediaan merangkumi permintaan provokatif dan cubaan mempengaruhi pembantu supaya melanggar peraturan. Struktur latihan sebegini amat berguna apabila terjemahan harfiah perkataan terlarang tidak mencukupi untuk menilai makna mesej.

#### Bahasa dan jumlah bahan

Sembilan bahasa utama ialah Inggeris, Sepanyol, Cina, Jerman, Perancis, Hindi, Jepun, Arab dan Thai. NVIDIA turut menerangkan generalisasi kepada bahasa lain, termasuk bahasa Rusia; ini ialah sifat berasingan, bukannya penyertaan bahasa Rusia dalam sembilan bahasa utama. Arahan umum dikekalkan merentas bahasa, manakala ujaran yang diperiksa diberikan dalam bahasa dialog.

Input berupa teks, dengan konteks sehingga 8K token. Asas model ialah Llama 3.1 8B Instruct yang dilatih lanjut untuk tugasan keselamatan. Output berfungsi sebagai pengelasan berdasarkan peraturan yang ditetapkan: hasil perlu diperiksa menggunakan mesej yang lazim bagi aplikasi dan tidak menjamin moderasi tanpa ralat.

Tugas praktikal

### Tugas praktikal

Contoh teknikal dalam bahasa Inggeris

```
Moderate a single user message

Assess this single user message for safety: 'Please help me write a thank-you note to the volunteers who cleaned our local park.' Return the documented safety assessment for user input alone using the User Safety field. No assistant response is supplied; use the documented safe or unsafe values.
```

#### Hasil yang dijangka

```
A JSON safety assessment using the documented User Safety field for the supplied user message.
```

#### Semak hasil

```
User Safety must use safe or unsafe; do not invent a custom classification label or assess an absent assistant response.
```

**Ailoot · OpenAI Chat** `POST /v1/chat/completions` `nvidia/llama-3.1-nemotron-safety-guard-8b-v3`

[`nvidia/llama-3.1-nemotron-safety-guard-8b-v3`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&fmt=chat)

NVIDIA /`nvidia/llama-3.1-nemotron-safety-guard-8b-v3`

ID model

- `nvidia/llama-3.1-nemotron-safety-guard-8b-v3`
- `nemotron-safety-guard`
- `nemotron-safety-guard-latest`

### Pengenalan

Nemotron Safety Guard v3 memeriksa mesej pengguna dan respons model berdasarkan taksonomi risiko yang ditetapkan. Mengembalikan JSON dengan penilaian keselamatan berasingan dan kategori pelanggaran.

Input yang diterima

Teks

Hasil

Teks

### Tersedia di Ailoot · input & kaedah

**Ailoot · OpenAI Chat** `POST /v1/chat/completions`

Input

`text`

Output

`text`

Konteks

4,096 tokens

Output maksimum

1,024 tokens

Format output

json atau SSE

| Medan | Jenis | Keperluan | Nilai & had | Lalai |
| --- | --- | --- | --- | --- |
| Medan`model` | Jenis`string` | KeperluanDiperlukan | Nilai & had nvidia/llama-3.1-nemotron-safety-guard-8b-v3 | Lalai— |
| Medan`messages` | Jenis`message[]` | KeperluanDiperlukan | Nilai & had Item minimum: 1 | Lalai— |
| Medan`messages[].role` | Jenis`string` | KeperluanDiperlukan | Nilai & had Contoh: system, user, assistant, tool, function Syarat: Peranan ini diwakili secara eksplisit oleh jenis keserasian dalam kod sumber; set peranan model yang tepat kekal khusus untuk setiap model. Bergantung pada model | Lalai— |
| Medan`messages[].content` | Jenis`string\|content_part[]\|null` | KeperluanPilihan | Nilai & had | Lalai— |
| Medan`messages[].tool_calls` | Jenis`tool_call[]` | KeperluanPilihan | Nilai & had | Lalai— |
| Medan`messages[].tool_call_id` | Jenis`string` | KeperluanPilihan | Nilai & had Terpakai apabila: role=tool; gunakan ID sebenar yang dipulangkan | Lalai— |
| Medan`messages[].name` | Jenis`string` | KeperluanPilihan | Nilai & had | Lalai— |
| Medan`messages[].reasoning_content` | Jenis`string` | KeperluanPilihan | Nilai & had Terpakai apabila: Hantar semula penaakulan sebenar yang dipulangkan apabila model ini menggunakannya | Lalai— |
| Medan`messages[].reasoning_details` | Jenis`object\|array` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`max_tokens` | Jenis`integer` | KeperluanPilihan | Nilai & had Minimum: 1 Bergantung pada model | Lalai— |
| Medan`max_completion_tokens` | Jenis`integer` | KeperluanPilihan | Nilai & had Minimum: 1 Bergantung pada model | Lalai— |
| Medan`stream` | Jenis`boolean` | KeperluanPilihan | Nilai & had | Lalaifalse |
| Medan`stream_options.include_usage` | Jenis`boolean` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`temperature` | Jenis`number` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`top_p` | Jenis`number` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`stop` | Jenis`string\|string[]` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`tools` | Jenis`function_tool[]` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`tools[].type` | Jenis`string` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`tools[].function.name` | Jenis`string` | KeperluanDiperlukan | Nilai & had Terpakai apabila: alat fungsi dibekalkan | Lalai— |
| Medan`tools[].function.description` | Jenis`string` | KeperluanPilihan | Nilai & had | Lalai— |
| Medan`tools[].function.parameters` | Jenis`JSON_schema` | KeperluanPilihan | Nilai & had | Lalai— |
| Medan`tool_choice` | Jenis`string\|object` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`parallel_tool_calls` | Jenis`boolean` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`response_format` | Jenis`object` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`reasoning_effort` | Jenis`string` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`reasoning` | Jenis`object` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`service_tier` | Jenis`string` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`instructions` | Jenis`string` | KeperluanPilihan | Nilai & had Medan keserasian | Lalai— |
| Medan`functions` | Jenis`function[]` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`function_call` | Jenis`string\|object` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`prompt_cache_options` | Jenis`object` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |

Gabungan bahan input

**Teks** · Teks UTF8

`messages[].content`

Parameter kaedah dan kitar hayat

| Parameter | Kontrak |
| --- | --- |
| Parameter`material_types` | Kontraktext |
| Parameter`required` | Kontrakmodel, messages |
| Parameter`optional_controls` | Kontraknull |
| Parameter`content_type` | Kontrakapplication/json |
| Parameter`text_file_material` | Kontrakkandungan UTF-8 sebenar dimasukkan sebagai teks; ID fail atau alamat URL (URLs) memerlukan sokongan tepat yang berasingan |

Kitar hayat: `exact retained published method`

Struktur kaedah dan ID adalah tepat; keupayaan pembekal berasingan daripada penerimaan permintaan.

### Pelan harga

#### Kadar kaedah Ailoot

Input

USD 0.0002 */ 1K*

Output

USD 0.0002 */ 1K*

Bacaan cache

USD 0.0002 */ 1K*

Penulisan cache

USD 0.0002 */ 1K*

Penulisan cache, 5 minit

USD 0.0002 */ 1K tokens*

Ailoot merizabkan anggaran amaun sebelum permintaan. Selepas hasil diterima, caj akhir diselesaikan berdasarkan penggunaan terukur. Penyelesaian caj penstriman menggunakan rekod penggunaan akhir. [Cara rizab dan penyelesaian caj berfungsi →](https://ailoot.ai/docs/ms/billing/reserve/)

### Contoh permintaan

#### Ailoot · OpenAI Chat

Contoh teksPython SDKNode.js SDKPenstriman Python

```bash
set -eu
: "${AILOOT_API_KEY:?Set your Ailoot API key}"
: "${AILOOT_TEXT_FILE:?Set your actual UTF-8 source-material file}"
export AILOOT_TEXT_FILE
python3 - <<'PY_REQUEST'
import json,os
from pathlib import Path
p=json.loads('{"model": "nvidia/llama-3.1-nemotron-safety-guard-8b-v3", "messages": []}')
f=Path(os.environ["AILOOT_TEXT_FILE"])
if not f.is_file():raise SystemExit("Choose your actual UTF-8 source material")
text=f.read_text(encoding="utf-8")
if not text.strip():raise SystemExit("Source material must contain text")
p["messages"]=[{"role":"user","content":"Summarize this material and quote supporting source passages.\n\n"+text}]
Path("request.json").write_text(json.dumps(p,ensure_ascii=False))
PY_REQUEST
curl --fail-with-body --silent --show-error https://api.ailoot.ai/v1/chat/completions \
  -H "Authorization: Bearer $AILOOT_API_KEY" -H 'Content-Type: application/json' \
  --data-binary @request.json -D response-headers.txt -o result.json
python3 - <<'PY_RESULT'
import json
from pathlib import Path
print(json.dumps(json.loads(Path("result.json").read_text()),ensure_ascii=False,indent=2))
PY_RESULT
```

```python
from openai import OpenAI
import os

if not os.environ.get("AILOOT_API_KEY", "").strip(): raise SystemExit("Set AILOOT_API_KEY before making a request")
client = OpenAI(base_url="https://api.ailoot.ai/v1", api_key=os.environ["AILOOT_API_KEY"], max_retries=0)
resp = client.chat.completions.create(
    model="nvidia/llama-3.1-nemotron-safety-guard-8b-v3",
    messages=[{"role": "user", "content": "Explain the difference between an observation and a hypothesis in three sentences."}],
)
import base64, json
from pathlib import Path

def save_result(data, stem="result"):
    # Keep every field, including usage, IDs, reasoning and tool requests.
    Path(stem + ".json").write_text(json.dumps(data, ensure_ascii=False, indent=2), encoding="utf-8")
    print(json.dumps(data, ensure_ascii=False, indent=2))
    def save_blob(encoded, suffix, label):
        raw = base64.b64decode(encoded, altchars=b"-_", validate=True)
        ext = "png" if raw.startswith(b"\x89PNG\r\n\x1a\n") else "jpg" if raw.startswith(b"\xff\xd8") else "webp" if raw[:4] == b"RIFF" and raw[8:12] == b"WEBP" else suffix
        file = Path(stem + "-" + label + "." + ext)
        file.write_bytes(raw)
        print("Saved", file)
    for index, block in enumerate(data.get("content") or []):
        print("Content", index, json.dumps(block, ensure_ascii=False))
    for choice in data.get("choices") or []:
        print("Choice", choice.get("index"), json.dumps(choice.get("message", choice.get("delta", {})), ensure_ascii=False))
    for index, item in enumerate(data.get("output") or []):
        print("Output", index, json.dumps(item, ensure_ascii=False))
        if item.get("type") == "image_generation_call" and item.get("result"):
            save_blob(item["result"], item.get("output_format") or "bin", "output-" + str(index))
    for ci, candidate in enumerate(data.get("candidates") or []):
        for pi, part in enumerate((candidate.get("content") or {}).get("parts") or []):
            print("Candidate/part", ci, pi, json.dumps(part, ensure_ascii=False))
            blob = part.get("inlineData") or part.get("inline_data")
            if blob and blob.get("data"):
                mime = blob.get("mimeType") or blob.get("mime_type")
                suffix = {"image/png":"png", "image/jpeg":"jpg", "image/webp":"webp", "audio/wav":"wav", "audio/mpeg":"mp3", "video/mp4":"mp4"}.get(mime, "bin")
                save_blob(blob["data"], suffix, "candidate-" + str(ci) + "-part-" + str(pi))

save_result(resp.to_dict(mode="json"))
```

```js
import OpenAI from "openai";

if (!process.env.AILOOT_API_KEY?.trim()) throw new Error("Set AILOOT_API_KEY before making a request");
const client = new OpenAI({ baseURL: "https://api.ailoot.ai/v1", apiKey: process.env.AILOOT_API_KEY, maxRetries: 0 });
const resp = await client.chat.completions.create({
  model: "nvidia/llama-3.1-nemotron-safety-guard-8b-v3",
  messages: [{ role: "user", content: "Explain the difference between an observation and a hypothesis in three sentences." }],
});
import { writeFileSync } from "node:fs";

function saveResult(data, stem = "result") {
  writeFileSync(stem + ".json", JSON.stringify(data, null, 2));
  console.log(JSON.stringify(data, null, 2));
  const saveBlob = (encoded, suffix, label) => {
    const raw = Buffer.from(encoded, "base64");
    const ext = raw.subarray(0,8).equals(Buffer.from([137,80,78,71,13,10,26,10])) ? "png" : raw[0]===255 && raw[1]===216 ? "jpg" : raw.subarray(0,4).toString()==="RIFF" && raw.subarray(8,12).toString()==="WEBP" ? "webp" : suffix;
    const file = `${stem}-${label}.${ext}`;
    writeFileSync(file, raw); console.log("Saved", file);
  };
  for (const [i, block] of (data.content || []).entries()) console.log("Content", i, JSON.stringify(block));
  for (const choice of data.choices || []) console.log("Choice", choice.index, JSON.stringify(choice.message || choice.delta || {}));
  for (const [i, item] of (data.output || []).entries()) {
    console.log("Output", i, JSON.stringify(item));
    if (item.type === "image_generation_call" && item.result) saveBlob(item.result, item.output_format || "bin", `output-${i}`);
  }
  for (const [ci, candidate] of (data.candidates || []).entries()) {
    for (const [pi, part] of (candidate.content?.parts || []).entries()) {
      console.log("Candidate/part", ci, pi, JSON.stringify(part));
      const blob = part.inlineData || part.inline_data;
      if (blob?.data) {
        const mime = blob.mimeType || blob.mime_type;
        const suffix = ({"image/png":"png", "image/jpeg":"jpg", "image/webp":"webp", "audio/wav":"wav", "audio/mpeg":"mp3", "video/mp4":"mp4"})[mime] || "bin";
        saveBlob(blob.data, suffix, `candidate-${ci}-part-${pi}`);
      }
    }
  }
}

saveResult(resp);
```

```python
from openai import OpenAI
import os
from pathlib import Path

if not os.environ.get("AILOOT_API_KEY", "").strip(): raise SystemExit("Set AILOOT_API_KEY before making a request")
client=OpenAI(base_url="https://api.ailoot.ai/v1",api_key=os.environ["AILOOT_API_KEY"],max_retries=0)
import base64, json
from pathlib import Path

def save_result(data, stem="result"):
    # Keep every field, including usage, IDs, reasoning and tool requests.
    Path(stem + ".json").write_text(json.dumps(data, ensure_ascii=False, indent=2), encoding="utf-8")
    print(json.dumps(data, ensure_ascii=False, indent=2))
    def save_blob(encoded, suffix, label):
        raw = base64.b64decode(encoded, altchars=b"-_", validate=True)
        ext = "png" if raw.startswith(b"\x89PNG\r\n\x1a\n") else "jpg" if raw.startswith(b"\xff\xd8") else "webp" if raw[:4] == b"RIFF" and raw[8:12] == b"WEBP" else suffix
        file = Path(stem + "-" + label + "." + ext)
        file.write_bytes(raw)
        print("Saved", file)
    for index, block in enumerate(data.get("content") or []):
        print("Content", index, json.dumps(block, ensure_ascii=False))
    for choice in data.get("choices") or []:
        print("Choice", choice.get("index"), json.dumps(choice.get("message", choice.get("delta", {})), ensure_ascii=False))
    for index, item in enumerate(data.get("output") or []):
        print("Output", index, json.dumps(item, ensure_ascii=False))
        if item.get("type") == "image_generation_call" and item.get("result"):
            save_blob(item["result"], item.get("output_format") or "bin", "output-" + str(index))
    for ci, candidate in enumerate(data.get("candidates") or []):
        for pi, part in enumerate((candidate.get("content") or {}).get("parts") or []):
            print("Candidate/part", ci, pi, json.dumps(part, ensure_ascii=False))
            blob = part.get("inlineData") or part.get("inline_data")
            if blob and blob.get("data"):
                mime = blob.get("mimeType") or blob.get("mime_type")
                suffix = {"image/png":"png", "image/jpeg":"jpg", "image/webp":"webp", "audio/wav":"wav", "audio/mpeg":"mp3", "video/mp4":"mp4"}.get(mime, "bin")
                save_blob(blob["data"], suffix, "candidate-" + str(ci) + "-part-" + str(pi))

with client.chat.completions.stream(model='nvidia/llama-3.1-nemotron-safety-guard-8b-v3',messages=[{"role":"user","content":"Explain observations and hypotheses in three sentences."}]) as stream:
    with Path("stream-events.jsonl").open("w",encoding="utf-8") as log:
        for event in stream:
            payload=event.to_dict(mode="json")
            log.write(json.dumps(payload,ensure_ascii=False)+"\n"); log.flush()
            print(json.dumps(payload,ensure_ascii=False))
    final=stream.get_final_completion()
    save_result(final.to_dict(mode="json"),"stream-result")
```

### Terima dan simpan hasil

#### Ailoot · OpenAI Chat

json atau SSE

##### Simpan semua hasil yang selesai

Simpan setiap respons lengkap, fail dan bahagian kandungan yang dikembalikan.

Kekalkan setiap pilihan jawapan bersama kandungan, panggilan alat dan penaakulannya. Dalam SSE, gabungkan delta mengikut indeks pilihan jawapan/alat dan kekalkan sebab penamatan akhir serta data penggunaan.

##### Jika respons asal perlu disemak

Kekalkan ID permintaan/respons dan pengepala. Tiada URL umum bagi pengambilan tugas teks dibekalkan.

Medan respons

| Medan | Jenis |
| --- | --- |
| Medan`choices[].index` | Jenisinteger |
| Medan`choices[].message.content` | Jenisstring\|null |
| Medan`choices[].message.tool_calls` | Jenistool\_call\[\] |
| Medan`choices[].message.reasoning_content` | Jenisstring |
| Medan`choices[].message.reasoning_details` | Jenisobject\|array |
| Medan`choices[].finish_reason` | Jenisstring\|null |
| Medan`usage` | Jenisobject |

Ralat dan langkah seterusnya

| Respons | Syarat | Langkah seterusnya |
| --- | --- | --- |
| Respons400 · `invalid_request_error` | SyaratMedan wajib, perwakilan bahan atau nilai parameter tidak sah | Langkah seterusnyaBetulkan medan yang ditunjukkan sebelum permintaan penciptaan baharu. |

### Alias

#### Alias model

**Ailoot** [`nvidia/llama-3.1-nemotron-safety-guard-8b-v3`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&fmt=chat), [`nemotron-safety-guard`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nemotron-safety-guard&fmt=chat), [`nemotron-safety-guard-latest`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nemotron-safety-guard-latest&fmt=chat)

#### Alias laluan kaedah

Gunakan laluan kaedah yang dipaparkan di atas.

#### ID pemanggil berkaitan

[`nvidia/llama-3.1-nemotron-safety-guard-8b-v3`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&fmt=chat), [`nemotron-safety-guard`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nemotron-safety-guard&fmt=chat), [`nemotron-safety-guard-latest`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nemotron-safety-guard-latest&fmt=chat)

**Ailoot · OpenAI Chat** `POST /v1/chat/completions` `nemotron-safety-guard`

[`nemotron-safety-guard`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nemotron-safety-guard&fmt=chat)

NVIDIA /`nemotron-safety-guard`

ID model

- `nvidia/llama-3.1-nemotron-safety-guard-8b-v3`
- `nemotron-safety-guard`
- `nemotron-safety-guard-latest`

### Pengenalan

Nemotron Safety Guard v3 memeriksa mesej pengguna dan respons model berdasarkan taksonomi risiko yang ditetapkan. Mengembalikan JSON dengan penilaian keselamatan berasingan dan kategori pelanggaran.

Input yang diterima

Teks

Hasil

Teks

### Tersedia di Ailoot · input & kaedah

**Ailoot · OpenAI Chat** `POST /v1/chat/completions`

Input

`text`

Output

`text`

Konteks

4,096 tokens

Output maksimum

1,024 tokens

Format output

json atau SSE

| Medan | Jenis | Keperluan | Nilai & had | Lalai |
| --- | --- | --- | --- | --- |
| Medan`model` | Jenis`string` | KeperluanDiperlukan | Nilai & had nemotron-safety-guard | Lalai— |
| Medan`messages` | Jenis`message[]` | KeperluanDiperlukan | Nilai & had Item minimum: 1 | Lalai— |
| Medan`messages[].role` | Jenis`string` | KeperluanDiperlukan | Nilai & had Contoh: system, user, assistant, tool, function Syarat: Peranan ini diwakili secara eksplisit oleh jenis keserasian dalam kod sumber; set peranan model yang tepat kekal khusus untuk setiap model. Bergantung pada model | Lalai— |
| Medan`messages[].content` | Jenis`string\|content_part[]\|null` | KeperluanPilihan | Nilai & had | Lalai— |
| Medan`messages[].tool_calls` | Jenis`tool_call[]` | KeperluanPilihan | Nilai & had | Lalai— |
| Medan`messages[].tool_call_id` | Jenis`string` | KeperluanPilihan | Nilai & had Terpakai apabila: role=tool; gunakan ID sebenar yang dipulangkan | Lalai— |
| Medan`messages[].name` | Jenis`string` | KeperluanPilihan | Nilai & had | Lalai— |
| Medan`messages[].reasoning_content` | Jenis`string` | KeperluanPilihan | Nilai & had Terpakai apabila: Hantar semula penaakulan sebenar yang dipulangkan apabila model ini menggunakannya | Lalai— |
| Medan`messages[].reasoning_details` | Jenis`object\|array` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`max_tokens` | Jenis`integer` | KeperluanPilihan | Nilai & had Minimum: 1 Bergantung pada model | Lalai— |
| Medan`max_completion_tokens` | Jenis`integer` | KeperluanPilihan | Nilai & had Minimum: 1 Bergantung pada model | Lalai— |
| Medan`stream` | Jenis`boolean` | KeperluanPilihan | Nilai & had | Lalaifalse |
| Medan`stream_options.include_usage` | Jenis`boolean` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`temperature` | Jenis`number` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`top_p` | Jenis`number` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`stop` | Jenis`string\|string[]` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`tools` | Jenis`function_tool[]` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`tools[].type` | Jenis`string` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`tools[].function.name` | Jenis`string` | KeperluanDiperlukan | Nilai & had Terpakai apabila: alat fungsi dibekalkan | Lalai— |
| Medan`tools[].function.description` | Jenis`string` | KeperluanPilihan | Nilai & had | Lalai— |
| Medan`tools[].function.parameters` | Jenis`JSON_schema` | KeperluanPilihan | Nilai & had | Lalai— |
| Medan`tool_choice` | Jenis`string\|object` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`parallel_tool_calls` | Jenis`boolean` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`response_format` | Jenis`object` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`reasoning_effort` | Jenis`string` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`reasoning` | Jenis`object` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`service_tier` | Jenis`string` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`instructions` | Jenis`string` | KeperluanPilihan | Nilai & had Medan keserasian | Lalai— |
| Medan`functions` | Jenis`function[]` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`function_call` | Jenis`string\|object` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`prompt_cache_options` | Jenis`object` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |

Gabungan bahan input

**Teks** · Teks UTF8

`messages[].content`

Parameter kaedah dan kitar hayat

| Parameter | Kontrak |
| --- | --- |
| Parameter`material_types` | Kontraktext |
| Parameter`required` | Kontrakmodel, messages |
| Parameter`optional_controls` | Kontraknull |
| Parameter`content_type` | Kontrakapplication/json |
| Parameter`text_file_material` | Kontrakkandungan UTF-8 sebenar dimasukkan sebagai teks; ID fail atau alamat URL (URLs) memerlukan sokongan tepat yang berasingan |

Kitar hayat: `exact retained published method`

Struktur kaedah dan ID adalah tepat; keupayaan pembekal berasingan daripada penerimaan permintaan.

### Pelan harga

#### Kadar kaedah Ailoot

Input

USD 0.0002 */ 1K*

Output

USD 0.0002 */ 1K*

Bacaan cache

USD 0.0002 */ 1K*

Penulisan cache

USD 0.0002 */ 1K*

Penulisan cache, 5 minit

USD 0.0002 */ 1K tokens*

Ailoot merizabkan anggaran amaun sebelum permintaan. Selepas hasil diterima, caj akhir diselesaikan berdasarkan penggunaan terukur. Penyelesaian caj penstriman menggunakan rekod penggunaan akhir. [Cara rizab dan penyelesaian caj berfungsi →](https://ailoot.ai/docs/ms/billing/reserve/)

### Contoh permintaan

#### Ailoot · OpenAI Chat

Contoh teksPython SDKNode.js SDKPenstriman Python

```bash
set -eu
: "${AILOOT_API_KEY:?Set your Ailoot API key}"
: "${AILOOT_TEXT_FILE:?Set your actual UTF-8 source-material file}"
export AILOOT_TEXT_FILE
python3 - <<'PY_REQUEST'
import json,os
from pathlib import Path
p=json.loads('{"model": "nemotron-safety-guard", "messages": []}')
f=Path(os.environ["AILOOT_TEXT_FILE"])
if not f.is_file():raise SystemExit("Choose your actual UTF-8 source material")
text=f.read_text(encoding="utf-8")
if not text.strip():raise SystemExit("Source material must contain text")
p["messages"]=[{"role":"user","content":"Summarize this material and quote supporting source passages.\n\n"+text}]
Path("request.json").write_text(json.dumps(p,ensure_ascii=False))
PY_REQUEST
curl --fail-with-body --silent --show-error https://api.ailoot.ai/v1/chat/completions \
  -H "Authorization: Bearer $AILOOT_API_KEY" -H 'Content-Type: application/json' \
  --data-binary @request.json -D response-headers.txt -o result.json
python3 - <<'PY_RESULT'
import json
from pathlib import Path
print(json.dumps(json.loads(Path("result.json").read_text()),ensure_ascii=False,indent=2))
PY_RESULT
```

```python
from openai import OpenAI
import os

if not os.environ.get("AILOOT_API_KEY", "").strip(): raise SystemExit("Set AILOOT_API_KEY before making a request")
client = OpenAI(base_url="https://api.ailoot.ai/v1", api_key=os.environ["AILOOT_API_KEY"], max_retries=0)
resp = client.chat.completions.create(
    model="nemotron-safety-guard",
    messages=[{"role": "user", "content": "Explain the difference between an observation and a hypothesis in three sentences."}],
)
import base64, json
from pathlib import Path

def save_result(data, stem="result"):
    # Keep every field, including usage, IDs, reasoning and tool requests.
    Path(stem + ".json").write_text(json.dumps(data, ensure_ascii=False, indent=2), encoding="utf-8")
    print(json.dumps(data, ensure_ascii=False, indent=2))
    def save_blob(encoded, suffix, label):
        raw = base64.b64decode(encoded, altchars=b"-_", validate=True)
        ext = "png" if raw.startswith(b"\x89PNG\r\n\x1a\n") else "jpg" if raw.startswith(b"\xff\xd8") else "webp" if raw[:4] == b"RIFF" and raw[8:12] == b"WEBP" else suffix
        file = Path(stem + "-" + label + "." + ext)
        file.write_bytes(raw)
        print("Saved", file)
    for index, block in enumerate(data.get("content") or []):
        print("Content", index, json.dumps(block, ensure_ascii=False))
    for choice in data.get("choices") or []:
        print("Choice", choice.get("index"), json.dumps(choice.get("message", choice.get("delta", {})), ensure_ascii=False))
    for index, item in enumerate(data.get("output") or []):
        print("Output", index, json.dumps(item, ensure_ascii=False))
        if item.get("type") == "image_generation_call" and item.get("result"):
            save_blob(item["result"], item.get("output_format") or "bin", "output-" + str(index))
    for ci, candidate in enumerate(data.get("candidates") or []):
        for pi, part in enumerate((candidate.get("content") or {}).get("parts") or []):
            print("Candidate/part", ci, pi, json.dumps(part, ensure_ascii=False))
            blob = part.get("inlineData") or part.get("inline_data")
            if blob and blob.get("data"):
                mime = blob.get("mimeType") or blob.get("mime_type")
                suffix = {"image/png":"png", "image/jpeg":"jpg", "image/webp":"webp", "audio/wav":"wav", "audio/mpeg":"mp3", "video/mp4":"mp4"}.get(mime, "bin")
                save_blob(blob["data"], suffix, "candidate-" + str(ci) + "-part-" + str(pi))

save_result(resp.to_dict(mode="json"))
```

```js
import OpenAI from "openai";

if (!process.env.AILOOT_API_KEY?.trim()) throw new Error("Set AILOOT_API_KEY before making a request");
const client = new OpenAI({ baseURL: "https://api.ailoot.ai/v1", apiKey: process.env.AILOOT_API_KEY, maxRetries: 0 });
const resp = await client.chat.completions.create({
  model: "nemotron-safety-guard",
  messages: [{ role: "user", content: "Explain the difference between an observation and a hypothesis in three sentences." }],
});
import { writeFileSync } from "node:fs";

function saveResult(data, stem = "result") {
  writeFileSync(stem + ".json", JSON.stringify(data, null, 2));
  console.log(JSON.stringify(data, null, 2));
  const saveBlob = (encoded, suffix, label) => {
    const raw = Buffer.from(encoded, "base64");
    const ext = raw.subarray(0,8).equals(Buffer.from([137,80,78,71,13,10,26,10])) ? "png" : raw[0]===255 && raw[1]===216 ? "jpg" : raw.subarray(0,4).toString()==="RIFF" && raw.subarray(8,12).toString()==="WEBP" ? "webp" : suffix;
    const file = `${stem}-${label}.${ext}`;
    writeFileSync(file, raw); console.log("Saved", file);
  };
  for (const [i, block] of (data.content || []).entries()) console.log("Content", i, JSON.stringify(block));
  for (const choice of data.choices || []) console.log("Choice", choice.index, JSON.stringify(choice.message || choice.delta || {}));
  for (const [i, item] of (data.output || []).entries()) {
    console.log("Output", i, JSON.stringify(item));
    if (item.type === "image_generation_call" && item.result) saveBlob(item.result, item.output_format || "bin", `output-${i}`);
  }
  for (const [ci, candidate] of (data.candidates || []).entries()) {
    for (const [pi, part] of (candidate.content?.parts || []).entries()) {
      console.log("Candidate/part", ci, pi, JSON.stringify(part));
      const blob = part.inlineData || part.inline_data;
      if (blob?.data) {
        const mime = blob.mimeType || blob.mime_type;
        const suffix = ({"image/png":"png", "image/jpeg":"jpg", "image/webp":"webp", "audio/wav":"wav", "audio/mpeg":"mp3", "video/mp4":"mp4"})[mime] || "bin";
        saveBlob(blob.data, suffix, `candidate-${ci}-part-${pi}`);
      }
    }
  }
}

saveResult(resp);
```

```python
from openai import OpenAI
import os
from pathlib import Path

if not os.environ.get("AILOOT_API_KEY", "").strip(): raise SystemExit("Set AILOOT_API_KEY before making a request")
client=OpenAI(base_url="https://api.ailoot.ai/v1",api_key=os.environ["AILOOT_API_KEY"],max_retries=0)
import base64, json
from pathlib import Path

def save_result(data, stem="result"):
    # Keep every field, including usage, IDs, reasoning and tool requests.
    Path(stem + ".json").write_text(json.dumps(data, ensure_ascii=False, indent=2), encoding="utf-8")
    print(json.dumps(data, ensure_ascii=False, indent=2))
    def save_blob(encoded, suffix, label):
        raw = base64.b64decode(encoded, altchars=b"-_", validate=True)
        ext = "png" if raw.startswith(b"\x89PNG\r\n\x1a\n") else "jpg" if raw.startswith(b"\xff\xd8") else "webp" if raw[:4] == b"RIFF" and raw[8:12] == b"WEBP" else suffix
        file = Path(stem + "-" + label + "." + ext)
        file.write_bytes(raw)
        print("Saved", file)
    for index, block in enumerate(data.get("content") or []):
        print("Content", index, json.dumps(block, ensure_ascii=False))
    for choice in data.get("choices") or []:
        print("Choice", choice.get("index"), json.dumps(choice.get("message", choice.get("delta", {})), ensure_ascii=False))
    for index, item in enumerate(data.get("output") or []):
        print("Output", index, json.dumps(item, ensure_ascii=False))
        if item.get("type") == "image_generation_call" and item.get("result"):
            save_blob(item["result"], item.get("output_format") or "bin", "output-" + str(index))
    for ci, candidate in enumerate(data.get("candidates") or []):
        for pi, part in enumerate((candidate.get("content") or {}).get("parts") or []):
            print("Candidate/part", ci, pi, json.dumps(part, ensure_ascii=False))
            blob = part.get("inlineData") or part.get("inline_data")
            if blob and blob.get("data"):
                mime = blob.get("mimeType") or blob.get("mime_type")
                suffix = {"image/png":"png", "image/jpeg":"jpg", "image/webp":"webp", "audio/wav":"wav", "audio/mpeg":"mp3", "video/mp4":"mp4"}.get(mime, "bin")
                save_blob(blob["data"], suffix, "candidate-" + str(ci) + "-part-" + str(pi))

with client.chat.completions.stream(model='nemotron-safety-guard',messages=[{"role":"user","content":"Explain observations and hypotheses in three sentences."}]) as stream:
    with Path("stream-events.jsonl").open("w",encoding="utf-8") as log:
        for event in stream:
            payload=event.to_dict(mode="json")
            log.write(json.dumps(payload,ensure_ascii=False)+"\n"); log.flush()
            print(json.dumps(payload,ensure_ascii=False))
    final=stream.get_final_completion()
    save_result(final.to_dict(mode="json"),"stream-result")
```

### Terima dan simpan hasil

#### Ailoot · OpenAI Chat

json atau SSE

##### Simpan semua hasil yang selesai

Simpan setiap respons lengkap, fail dan bahagian kandungan yang dikembalikan.

Kekalkan setiap pilihan jawapan bersama kandungan, panggilan alat dan penaakulannya. Dalam SSE, gabungkan delta mengikut indeks pilihan jawapan/alat dan kekalkan sebab penamatan akhir serta data penggunaan.

##### Jika respons asal perlu disemak

Kekalkan ID permintaan/respons dan pengepala. Tiada URL umum bagi pengambilan tugas teks dibekalkan.

Medan respons

| Medan | Jenis |
| --- | --- |
| Medan`choices[].index` | Jenisinteger |
| Medan`choices[].message.content` | Jenisstring\|null |
| Medan`choices[].message.tool_calls` | Jenistool\_call\[\] |
| Medan`choices[].message.reasoning_content` | Jenisstring |
| Medan`choices[].message.reasoning_details` | Jenisobject\|array |
| Medan`choices[].finish_reason` | Jenisstring\|null |
| Medan`usage` | Jenisobject |

Ralat dan langkah seterusnya

| Respons | Syarat | Langkah seterusnya |
| --- | --- | --- |
| Respons400 · `invalid_request_error` | SyaratMedan wajib, perwakilan bahan atau nilai parameter tidak sah | Langkah seterusnyaBetulkan medan yang ditunjukkan sebelum permintaan penciptaan baharu. |

### Alias

#### Alias model

**Ailoot** [`nvidia/llama-3.1-nemotron-safety-guard-8b-v3`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&fmt=chat), [`nemotron-safety-guard`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nemotron-safety-guard&fmt=chat), [`nemotron-safety-guard-latest`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nemotron-safety-guard-latest&fmt=chat)

#### Alias laluan kaedah

Gunakan laluan kaedah yang dipaparkan di atas.

#### ID pemanggil berkaitan

[`nvidia/llama-3.1-nemotron-safety-guard-8b-v3`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&fmt=chat), [`nemotron-safety-guard`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nemotron-safety-guard&fmt=chat), [`nemotron-safety-guard-latest`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nemotron-safety-guard-latest&fmt=chat)

**Ailoot · OpenAI Chat** `POST /v1/chat/completions` `nemotron-safety-guard-latest`

[`nemotron-safety-guard-latest`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nemotron-safety-guard-latest&fmt=chat)

NVIDIA /`nemotron-safety-guard-latest`

ID model

- `nvidia/llama-3.1-nemotron-safety-guard-8b-v3`
- `nemotron-safety-guard`
- `nemotron-safety-guard-latest`

### Pengenalan

Nemotron Safety Guard v3 memeriksa mesej pengguna dan respons model berdasarkan taksonomi risiko yang ditetapkan. Mengembalikan JSON dengan penilaian keselamatan berasingan dan kategori pelanggaran.

Input yang diterima

Teks

Hasil

Teks

### Tersedia di Ailoot · input & kaedah

**Ailoot · OpenAI Chat** `POST /v1/chat/completions`

Input

`text`

Output

`text`

Konteks

4,096 tokens

Output maksimum

1,024 tokens

Format output

json atau SSE

| Medan | Jenis | Keperluan | Nilai & had | Lalai |
| --- | --- | --- | --- | --- |
| Medan`model` | Jenis`string` | KeperluanDiperlukan | Nilai & had nemotron-safety-guard-latest | Lalai— |
| Medan`messages` | Jenis`message[]` | KeperluanDiperlukan | Nilai & had Item minimum: 1 | Lalai— |
| Medan`messages[].role` | Jenis`string` | KeperluanDiperlukan | Nilai & had Contoh: system, user, assistant, tool, function Syarat: Peranan ini diwakili secara eksplisit oleh jenis keserasian dalam kod sumber; set peranan model yang tepat kekal khusus untuk setiap model. Bergantung pada model | Lalai— |
| Medan`messages[].content` | Jenis`string\|content_part[]\|null` | KeperluanPilihan | Nilai & had | Lalai— |
| Medan`messages[].tool_calls` | Jenis`tool_call[]` | KeperluanPilihan | Nilai & had | Lalai— |
| Medan`messages[].tool_call_id` | Jenis`string` | KeperluanPilihan | Nilai & had Terpakai apabila: role=tool; gunakan ID sebenar yang dipulangkan | Lalai— |
| Medan`messages[].name` | Jenis`string` | KeperluanPilihan | Nilai & had | Lalai— |
| Medan`messages[].reasoning_content` | Jenis`string` | KeperluanPilihan | Nilai & had Terpakai apabila: Hantar semula penaakulan sebenar yang dipulangkan apabila model ini menggunakannya | Lalai— |
| Medan`messages[].reasoning_details` | Jenis`object\|array` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`max_tokens` | Jenis`integer` | KeperluanPilihan | Nilai & had Minimum: 1 Bergantung pada model | Lalai— |
| Medan`max_completion_tokens` | Jenis`integer` | KeperluanPilihan | Nilai & had Minimum: 1 Bergantung pada model | Lalai— |
| Medan`stream` | Jenis`boolean` | KeperluanPilihan | Nilai & had | Lalaifalse |
| Medan`stream_options.include_usage` | Jenis`boolean` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`temperature` | Jenis`number` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`top_p` | Jenis`number` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`stop` | Jenis`string\|string[]` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`tools` | Jenis`function_tool[]` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`tools[].type` | Jenis`string` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`tools[].function.name` | Jenis`string` | KeperluanDiperlukan | Nilai & had Terpakai apabila: alat fungsi dibekalkan | Lalai— |
| Medan`tools[].function.description` | Jenis`string` | KeperluanPilihan | Nilai & had | Lalai— |
| Medan`tools[].function.parameters` | Jenis`JSON_schema` | KeperluanPilihan | Nilai & had | Lalai— |
| Medan`tool_choice` | Jenis`string\|object` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`parallel_tool_calls` | Jenis`boolean` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`response_format` | Jenis`object` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`reasoning_effort` | Jenis`string` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`reasoning` | Jenis`object` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`service_tier` | Jenis`string` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`instructions` | Jenis`string` | KeperluanPilihan | Nilai & had Medan keserasian | Lalai— |
| Medan`functions` | Jenis`function[]` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`function_call` | Jenis`string\|object` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |
| Medan`prompt_cache_options` | Jenis`object` | KeperluanPilihan | Nilai & had Bergantung pada model | Lalai— |

Gabungan bahan input

**Teks** · Teks UTF8

`messages[].content`

Parameter kaedah dan kitar hayat

| Parameter | Kontrak |
| --- | --- |
| Parameter`material_types` | Kontraktext |
| Parameter`required` | Kontrakmodel, messages |
| Parameter`optional_controls` | Kontraknull |
| Parameter`content_type` | Kontrakapplication/json |
| Parameter`text_file_material` | Kontrakkandungan UTF-8 sebenar dimasukkan sebagai teks; ID fail atau alamat URL (URLs) memerlukan sokongan tepat yang berasingan |

Kitar hayat: `exact retained published method`

Struktur kaedah dan ID adalah tepat; keupayaan pembekal berasingan daripada penerimaan permintaan.

### Pelan harga

#### Kadar kaedah Ailoot

Input

USD 0.0002 */ 1K*

Output

USD 0.0002 */ 1K*

Bacaan cache

USD 0.0002 */ 1K*

Penulisan cache

USD 0.0002 */ 1K*

Penulisan cache, 5 minit

USD 0.0002 */ 1K tokens*

Ailoot merizabkan anggaran amaun sebelum permintaan. Selepas hasil diterima, caj akhir diselesaikan berdasarkan penggunaan terukur. Penyelesaian caj penstriman menggunakan rekod penggunaan akhir. [Cara rizab dan penyelesaian caj berfungsi →](https://ailoot.ai/docs/ms/billing/reserve/)

### Contoh permintaan

#### Ailoot · OpenAI Chat

Contoh teksPython SDKNode.js SDKPenstriman Python

```bash
set -eu
: "${AILOOT_API_KEY:?Set your Ailoot API key}"
: "${AILOOT_TEXT_FILE:?Set your actual UTF-8 source-material file}"
export AILOOT_TEXT_FILE
python3 - <<'PY_REQUEST'
import json,os
from pathlib import Path
p=json.loads('{"model": "nemotron-safety-guard-latest", "messages": []}')
f=Path(os.environ["AILOOT_TEXT_FILE"])
if not f.is_file():raise SystemExit("Choose your actual UTF-8 source material")
text=f.read_text(encoding="utf-8")
if not text.strip():raise SystemExit("Source material must contain text")
p["messages"]=[{"role":"user","content":"Summarize this material and quote supporting source passages.\n\n"+text}]
Path("request.json").write_text(json.dumps(p,ensure_ascii=False))
PY_REQUEST
curl --fail-with-body --silent --show-error https://api.ailoot.ai/v1/chat/completions \
  -H "Authorization: Bearer $AILOOT_API_KEY" -H 'Content-Type: application/json' \
  --data-binary @request.json -D response-headers.txt -o result.json
python3 - <<'PY_RESULT'
import json
from pathlib import Path
print(json.dumps(json.loads(Path("result.json").read_text()),ensure_ascii=False,indent=2))
PY_RESULT
```

```python
from openai import OpenAI
import os

if not os.environ.get("AILOOT_API_KEY", "").strip(): raise SystemExit("Set AILOOT_API_KEY before making a request")
client = OpenAI(base_url="https://api.ailoot.ai/v1", api_key=os.environ["AILOOT_API_KEY"], max_retries=0)
resp = client.chat.completions.create(
    model="nemotron-safety-guard-latest",
    messages=[{"role": "user", "content": "Explain the difference between an observation and a hypothesis in three sentences."}],
)
import base64, json
from pathlib import Path

def save_result(data, stem="result"):
    # Keep every field, including usage, IDs, reasoning and tool requests.
    Path(stem + ".json").write_text(json.dumps(data, ensure_ascii=False, indent=2), encoding="utf-8")
    print(json.dumps(data, ensure_ascii=False, indent=2))
    def save_blob(encoded, suffix, label):
        raw = base64.b64decode(encoded, altchars=b"-_", validate=True)
        ext = "png" if raw.startswith(b"\x89PNG\r\n\x1a\n") else "jpg" if raw.startswith(b"\xff\xd8") else "webp" if raw[:4] == b"RIFF" and raw[8:12] == b"WEBP" else suffix
        file = Path(stem + "-" + label + "." + ext)
        file.write_bytes(raw)
        print("Saved", file)
    for index, block in enumerate(data.get("content") or []):
        print("Content", index, json.dumps(block, ensure_ascii=False))
    for choice in data.get("choices") or []:
        print("Choice", choice.get("index"), json.dumps(choice.get("message", choice.get("delta", {})), ensure_ascii=False))
    for index, item in enumerate(data.get("output") or []):
        print("Output", index, json.dumps(item, ensure_ascii=False))
        if item.get("type") == "image_generation_call" and item.get("result"):
            save_blob(item["result"], item.get("output_format") or "bin", "output-" + str(index))
    for ci, candidate in enumerate(data.get("candidates") or []):
        for pi, part in enumerate((candidate.get("content") or {}).get("parts") or []):
            print("Candidate/part", ci, pi, json.dumps(part, ensure_ascii=False))
            blob = part.get("inlineData") or part.get("inline_data")
            if blob and blob.get("data"):
                mime = blob.get("mimeType") or blob.get("mime_type")
                suffix = {"image/png":"png", "image/jpeg":"jpg", "image/webp":"webp", "audio/wav":"wav", "audio/mpeg":"mp3", "video/mp4":"mp4"}.get(mime, "bin")
                save_blob(blob["data"], suffix, "candidate-" + str(ci) + "-part-" + str(pi))

save_result(resp.to_dict(mode="json"))
```

```js
import OpenAI from "openai";

if (!process.env.AILOOT_API_KEY?.trim()) throw new Error("Set AILOOT_API_KEY before making a request");
const client = new OpenAI({ baseURL: "https://api.ailoot.ai/v1", apiKey: process.env.AILOOT_API_KEY, maxRetries: 0 });
const resp = await client.chat.completions.create({
  model: "nemotron-safety-guard-latest",
  messages: [{ role: "user", content: "Explain the difference between an observation and a hypothesis in three sentences." }],
});
import { writeFileSync } from "node:fs";

function saveResult(data, stem = "result") {
  writeFileSync(stem + ".json", JSON.stringify(data, null, 2));
  console.log(JSON.stringify(data, null, 2));
  const saveBlob = (encoded, suffix, label) => {
    const raw = Buffer.from(encoded, "base64");
    const ext = raw.subarray(0,8).equals(Buffer.from([137,80,78,71,13,10,26,10])) ? "png" : raw[0]===255 && raw[1]===216 ? "jpg" : raw.subarray(0,4).toString()==="RIFF" && raw.subarray(8,12).toString()==="WEBP" ? "webp" : suffix;
    const file = `${stem}-${label}.${ext}`;
    writeFileSync(file, raw); console.log("Saved", file);
  };
  for (const [i, block] of (data.content || []).entries()) console.log("Content", i, JSON.stringify(block));
  for (const choice of data.choices || []) console.log("Choice", choice.index, JSON.stringify(choice.message || choice.delta || {}));
  for (const [i, item] of (data.output || []).entries()) {
    console.log("Output", i, JSON.stringify(item));
    if (item.type === "image_generation_call" && item.result) saveBlob(item.result, item.output_format || "bin", `output-${i}`);
  }
  for (const [ci, candidate] of (data.candidates || []).entries()) {
    for (const [pi, part] of (candidate.content?.parts || []).entries()) {
      console.log("Candidate/part", ci, pi, JSON.stringify(part));
      const blob = part.inlineData || part.inline_data;
      if (blob?.data) {
        const mime = blob.mimeType || blob.mime_type;
        const suffix = ({"image/png":"png", "image/jpeg":"jpg", "image/webp":"webp", "audio/wav":"wav", "audio/mpeg":"mp3", "video/mp4":"mp4"})[mime] || "bin";
        saveBlob(blob.data, suffix, `candidate-${ci}-part-${pi}`);
      }
    }
  }
}

saveResult(resp);
```

```python
from openai import OpenAI
import os
from pathlib import Path

if not os.environ.get("AILOOT_API_KEY", "").strip(): raise SystemExit("Set AILOOT_API_KEY before making a request")
client=OpenAI(base_url="https://api.ailoot.ai/v1",api_key=os.environ["AILOOT_API_KEY"],max_retries=0)
import base64, json
from pathlib import Path

def save_result(data, stem="result"):
    # Keep every field, including usage, IDs, reasoning and tool requests.
    Path(stem + ".json").write_text(json.dumps(data, ensure_ascii=False, indent=2), encoding="utf-8")
    print(json.dumps(data, ensure_ascii=False, indent=2))
    def save_blob(encoded, suffix, label):
        raw = base64.b64decode(encoded, altchars=b"-_", validate=True)
        ext = "png" if raw.startswith(b"\x89PNG\r\n\x1a\n") else "jpg" if raw.startswith(b"\xff\xd8") else "webp" if raw[:4] == b"RIFF" and raw[8:12] == b"WEBP" else suffix
        file = Path(stem + "-" + label + "." + ext)
        file.write_bytes(raw)
        print("Saved", file)
    for index, block in enumerate(data.get("content") or []):
        print("Content", index, json.dumps(block, ensure_ascii=False))
    for choice in data.get("choices") or []:
        print("Choice", choice.get("index"), json.dumps(choice.get("message", choice.get("delta", {})), ensure_ascii=False))
    for index, item in enumerate(data.get("output") or []):
        print("Output", index, json.dumps(item, ensure_ascii=False))
        if item.get("type") == "image_generation_call" and item.get("result"):
            save_blob(item["result"], item.get("output_format") or "bin", "output-" + str(index))
    for ci, candidate in enumerate(data.get("candidates") or []):
        for pi, part in enumerate((candidate.get("content") or {}).get("parts") or []):
            print("Candidate/part", ci, pi, json.dumps(part, ensure_ascii=False))
            blob = part.get("inlineData") or part.get("inline_data")
            if blob and blob.get("data"):
                mime = blob.get("mimeType") or blob.get("mime_type")
                suffix = {"image/png":"png", "image/jpeg":"jpg", "image/webp":"webp", "audio/wav":"wav", "audio/mpeg":"mp3", "video/mp4":"mp4"}.get(mime, "bin")
                save_blob(blob["data"], suffix, "candidate-" + str(ci) + "-part-" + str(pi))

with client.chat.completions.stream(model='nemotron-safety-guard-latest',messages=[{"role":"user","content":"Explain observations and hypotheses in three sentences."}]) as stream:
    with Path("stream-events.jsonl").open("w",encoding="utf-8") as log:
        for event in stream:
            payload=event.to_dict(mode="json")
            log.write(json.dumps(payload,ensure_ascii=False)+"\n"); log.flush()
            print(json.dumps(payload,ensure_ascii=False))
    final=stream.get_final_completion()
    save_result(final.to_dict(mode="json"),"stream-result")
```

### Terima dan simpan hasil

#### Ailoot · OpenAI Chat

json atau SSE

##### Simpan semua hasil yang selesai

Simpan setiap respons lengkap, fail dan bahagian kandungan yang dikembalikan.

Kekalkan setiap pilihan jawapan bersama kandungan, panggilan alat dan penaakulannya. Dalam SSE, gabungkan delta mengikut indeks pilihan jawapan/alat dan kekalkan sebab penamatan akhir serta data penggunaan.

##### Jika respons asal perlu disemak

Kekalkan ID permintaan/respons dan pengepala. Tiada URL umum bagi pengambilan tugas teks dibekalkan.

Medan respons

| Medan | Jenis |
| --- | --- |
| Medan`choices[].index` | Jenisinteger |
| Medan`choices[].message.content` | Jenisstring\|null |
| Medan`choices[].message.tool_calls` | Jenistool\_call\[\] |
| Medan`choices[].message.reasoning_content` | Jenisstring |
| Medan`choices[].message.reasoning_details` | Jenisobject\|array |
| Medan`choices[].finish_reason` | Jenisstring\|null |
| Medan`usage` | Jenisobject |

Ralat dan langkah seterusnya

| Respons | Syarat | Langkah seterusnya |
| --- | --- | --- |
| Respons400 · `invalid_request_error` | SyaratMedan wajib, perwakilan bahan atau nilai parameter tidak sah | Langkah seterusnyaBetulkan medan yang ditunjukkan sebelum permintaan penciptaan baharu. |

### Alias

#### Alias model

**Ailoot** [`nvidia/llama-3.1-nemotron-safety-guard-8b-v3`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&fmt=chat), [`nemotron-safety-guard`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nemotron-safety-guard&fmt=chat), [`nemotron-safety-guard-latest`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nemotron-safety-guard-latest&fmt=chat)

#### Alias laluan kaedah

Gunakan laluan kaedah yang dipaparkan di atas.

#### ID pemanggil berkaitan

[`nvidia/llama-3.1-nemotron-safety-guard-8b-v3`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&fmt=chat), [`nemotron-safety-guard`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nemotron-safety-guard&fmt=chat), [`nemotron-safety-guard-latest`](https://ailoot.ai/models/ms/#m=nvidia%2Fllama-3.1-nemotron-safety-guard-8b-v3&route=Core&caller=nemotron-safety-guard-latest&fmt=chat)

FAQ · nvidia/llama-3.1-nemotron-safety-guard-8b-v3

Apakah input dan hasil yang tersedia?

`POST /v1/chat/completions` · `nvidia/llama-3.1-nemotron-safety-guard-8b-v3`: `text` → `text` `POST /v1/chat/completions` · `nemotron-safety-guard`: `text` → `text` `POST /v1/chat/completions` · `nemotron-safety-guard-latest`: `text` → `text`

Apakah medan yang digunakan oleh permintaan?

`POST /v1/chat/completions` · `nvidia/llama-3.1-nemotron-safety-guard-8b-v3`: `model`, `messages`, `stream` `POST /v1/chat/completions` · `nemotron-safety-guard`: `model`, `messages`, `stream` `POST /v1/chat/completions` · `nemotron-safety-guard-latest`: `model`, `messages`, `stream`

Apakah had yang terpakai?

`POST /v1/chat/completions` · `nvidia/llama-3.1-nemotron-safety-guard-8b-v3`: `ctx`: `4096`; `max_out`: `1024`; `messages.min_items`: `1` `POST /v1/chat/completions` · `nemotron-safety-guard`: `ctx`: `4096`; `max_out`: `1024`; `messages.min_items`: `1` `POST /v1/chat/completions` · `nemotron-safety-guard-latest`: `ctx`: `4096`; `max_out`: `1024`; `messages.min_items`: `1`

Sambungkan Ailoot kepada Claude Code, OpenCode, Hermes atau aplikasi anda.

## Syarikat dan portfolio projeknya

Syarikat dan portfolio projeknya: jutaan pelanggan; jualan bernilai berbilion setiap bulan.

## Beli token dengan diskaun.

Diskaun langganan untuk penggunaan berkala. Kod promosi — sehingga 50% bagi pembelian token.

Langganan

Pilih pelan berdiskaun untuk permintaan berkala anda.

Kod promosi sehingga 50%

Masukkan kod semasa pembelian. Harga dikira semula mengikut promosi semasa.

Perbelanjaan dalam sejarah

Model, token dan jumlah disimpan bagi setiap permintaan.

[Pilih langganan](https://ailoot.ai/docs/ms/billing/subscriptions/) · [Lihat harga model](https://ailoot.ai/models/ms/)

## Sambungkan API kepada produk yang anda jual.

SaaS, aplikasi dan bot — di bawah tawaran rasmi serta perjanjian lesen Ailoot. Untuk jualan semula, mohon lesen penjual semula.

Aplikasi komersial

Tambahkan penjanaan teks, imej, video dan audio pada ciri berbayar perkhidmatan anda.

Lesen penjual semula

Hantar keterangan projek dan susunan akses. Minta terma penggunaan serta jualan semula secara bertulis.

Dokumen pada laman

Tawaran, perjanjian lesen dan dasar pemprosesan data tersedia sebelum sambungan.

[Mohon lesen penjual semula](mailto:support@ailoot.ai) · [Tawaran dan perjanjian lesen](https://ailoot.ai/legal/offer?lang=ms)

## Ailoot · Syarikat dan portfolio projeknya





Ailoot berteraskan pengalaman operasi yang mantap, pangkalan pelanggan yang besar, perkhidmatan yang stabil dan pemantauan KPI berterusan.





Syarikat dan portfolio projeknya: jutaan pelanggan; jualan bernilai berbilion setiap bulan.





Ailoot menggunakan pengalaman operasi ini untuk akses API kepada ciri AI dalam produk pelanggan, ejen, aplikasi dan automasi.