muapi Logomuapi Logo
muapi Logomuapi Logo

Gemma 4 31B SDFT Abliterated: AI Large Language Models

Usa Gemma 4 31B SDFT Abliterated tramite one API. 262K-token context, streaming e pay-per-token pricing. Provalo gratis nel playground Muapi. Gemma 4 31B SDFT Abliterated è un Gemma 4 31B variant con una 262K-token context window e image input, modified per reduce refusal behavior e tuned per creative chat; see also the Character Chat LLM API. Muapi exposes it tramite a single REST endpoint with async polling e Server-Sent Events streaming, so you can drop it into chat apps, agents e workflows without hosting GPUs. It è best suited per creative e character chat with image context. Pricing è pay-per-token at $0.612/M input tokens e $0.612/M token in output, senza abbonamento. Vedi anche Gemma 4 26B A4B Abliterated, Gemma 4 31B Gembrain Abliterated. Scopri l’intera gamma nella Gemma Abliterated API oppure nell’abliterated LLM API hub; famiglie correlate: GLM Abliterated API. 'Abliterated' e 'derestricted' describe modifications that reduce the model's refusal behavior. They do not guarantee unrestricted output, e you remain responsible per complying with applicable laws e the Muapi terms of use.

Controlli interattivi del modello

Usa Gemma 4 31B SDFT Abliterated tramite one API. 262K-token context, streaming e pay-per-token pricing. Provalo gratis nel playground Muapi.

📝

Panoramica

A proposito di questo modello

Gemma 4 31B SDFT Abliterated è un Gemma 4 31B variant con una 262K-token context window e image input, modified per reduce refusal behavior e tuned per creative chat; see also the Character Chat LLM API. Muapi exposes it tramite a single REST endpoint with async polling e Server-Sent Events streaming, so you can drop it into chat apps, agents e workflows without hosting GPUs. It è best suited per creative e character chat with image context. Pricing è pay-per-token at $0.612/M input tokens e $0.612/M token in output, senza abbonamento. Vedi anche Gemma 4 26B A4B Abliterated, Gemma 4 31B Gembrain Abliterated. Scopri l’intera gamma nella Gemma Abliterated API oppure nell’abliterated LLM API hub; famiglie correlate: GLM Abliterated API. 'Abliterated' e 'derestricted' describe modifications that reduce the model's refusal behavior. They do not guarantee unrestricted output, e you remain responsible per complying with applicable laws e the Muapi terms of use.

1Chat e assistants: Build conversational apps on Gemma 4 31B SDFT Abliterated per creative e character chat with image context.
2Sperimentazione con meno rifiuti: confronta il comportamento di rifiuto con il modello di base nelle tue valutazioni, nel rispetto delle norme d’uso Muapi.
3Agents e workflows: Usa the model da Muapi workflow nodes e the [abliterated LLM API](/abliterated-llm-api) hub examples.
💰

Prezzi e valore

Analisi dei costi

muapiapp$0.612/M input tokens and $0.612/M output tokens

Pay-per-token, billed on actual usage con una small minimum reserved up front.

Fal.aiNon disponibile

Gemma 4 31B SDFT Abliterated è not offered as a hosted endpoint.

ReplicateNon disponibile

Gemma 4 31B SDFT Abliterated è not offered as a hosted endpoint.

**I prezzi della concorrenza sono stimati in base ad architetture di modelli e livelli di utilizzo simili.

⚙️

Dettagli tecnici

Schema di configurazione

tempestivostring

.

Valore predefinitoSummarize the trade-offs between a monolith and microservices in five bullet points.
URL dell'immaginestring

Optional public image URL for multimodal analysis.

Valore predefinitoundefined
Richiesta di sistemastring

, .

Valore predefinitoYou are a concise, direct assistant.
Token massimiinteger

Optional maximum number of tokens to generate.

Valore predefinito8192
📖

Guida all'implementazione

Documentazione per gli sviluppatori

How per Use Gemma 4 31B SDFT Abliterated API

  1. Ottieni una chiave API: crea una chiave nella dashboard Muapi.

  2. Intramite un prompt: Invia POST prompt (plus optional system_prompt e max_tokens, up per 32768) per /api/v1/gemma-4-31b-sdft-abliterated. Provide an optional public image_url per include an image with your prompt.

curl -X POST https://api.muapi.ai/api/v1/gemma-4-31b-sdft-abliterated \
  -H "x-api-key: $MUAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{"prompt": "Explain how attention works in transformers.", "image_url": "https://example.com/photo.jpg"}'
  1. Esegui il polling o attiva lo streaming: Poll GET /api/v1/predictions/{request_id}/result until status è completed, or stream tokens live da /api/v1/gemma-4-31b-sdft-abliterated/stream.
curl -X POST https://api.muapi.ai/api/v1/gemma-4-31b-sdft-abliterated/stream \
  -H "x-api-key: $MUAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{"prompt": "Explain how attention works in transformers.", "image_url": "https://example.com/photo.jpg"}'
  1. Read the result: The generated text è returned in outputs. Usage è billed on actual token counts.
❓

Domande comuni

Chiesto frequentemente

What è Gemma 4 31B SDFT Abliterated?

Gemma 4 31B SDFT Abliterated è un Gemma 4 31B variant con una 262K-token context window e image input, modified per reduce refusal behavior e tuned per creative chat; see also the [Character Chat LLM API](/character-chat-llm-api).

What does abliterated mean?

Abliterated or derestricted models are modified per reduce refusal behavior. This lowers, but does not eliminate, refusals e does not guarantee unrestricted output.

How è Gemma 4 31B SDFT Abliterated priced?

Billing è per token: $0.612/M input tokens e $0.612/M output tokens. A small minimum è reserved when the request starts e reconciled per actual usage.

Does Gemma 4 31B SDFT Abliterated support streaming?

Yes. Stream tokens live da `/api/v1/gemma-4-31b-sdft-abliterated/stream` tramite Server-Sent Events oppure usa il polling dell’endpoint asincrono.

What è the context window?

The context window è about 262K tokens. Prompt e output together must fit inside it; `max_tokens` è capped at 32768. Image input è supported tramite `image_url`.

Is output guaranteed per be unrestricted?

No. Abliterated models are less likely per refuse, but behavior varies by prompt e quality can differ da the base model. Evaluate it per your use case e follow applicable laws.