muapi Logomuapi Logo
muapi Logomuapi Logo

Hermes 4 405B: AI Large Language Models

Call Hermes 4 405B through one API. 128K-token context, streaming and pay-per-token pricing. Try free in the Muapi playground. Hermes 4 405B is a 405B hybrid-reasoning chat model from Nous Research with a 128K-token context window, known for steerability and low refusal rates.

Interactive model controls

Hermes 4 405B is a 405B hybrid-reasoning chat model from Nous Research with a 128K-token context window, known for steerability and low refusal rates.

📝

Overview

About this model

Hermes 4 405B is a 405B hybrid-reasoning chat model from Nous Research with a 128K-token context window, known for steerability and low refusal rates. Muapi exposes it through a single REST endpoint with async polling and Server-Sent Events streaming, so you can drop it into chat apps, agents and workflows without hosting GPUs. It is best suited to steerable general chat, reasoning and agents. Pricing is pay-per-token at $0.6/M input tokens and $2.4/M output tokens, with no subscription. See also Venice Abliterated. Browse the full lineup on the Low-Refusal LLM API page or the Abliterated LLM API hub and the Character Chat LLM API hub. It is tuned to refuse less often than mainstream assistants, but it does not guarantee unrestricted output, and you remain responsible for complying with applicable laws and the Muapi terms of use.

1General chat: Build assistants on Hermes 4 405B for steerable general chat, reasoning and agents.
2Low-refusal experimentation: Compare refusal behavior against mainstream models in your own evaluations, within Muapi's usage policy.
3Agents and workflows: Call the model from Muapi workflow nodes and the [Low-Refusal LLM API](/low-refusal-llm-api) hub examples.
💰

Pricing & Value

Cost analysis

muapiapp$0.6/M input tokens and $2.4/M output tokens

Pay-per-token, billed on actual usage with a small minimum reserved up front.

Fal.aiNot available

Hermes 4 405B is not offered as a hosted endpoint.

ReplicateNot available

Hermes 4 405B is not offered as a hosted endpoint.

** Competitor pricing is estimated based on similar model architectures and usage tiers.

⚙️

Technical Details

Configuration schema

Promptstring

The user message or instruction for the model.

Default ValueSummarize the trade-offs between a monolith and microservices in five bullet points.
System Promptstring

Optional system-level instruction to guide model behavior.

Default ValueYou are a concise, direct assistant.
Max Tokensinteger

Optional maximum number of tokens to generate.

Default Value8192
📖

Implementation Guide

Developer documentation

How to Use Hermes 4 405B API

  1. Get an API key: Create a key in your Muapi dashboard.

  2. Send a prompt: POST prompt (plus optional system_prompt and max_tokens, up to 32768) to /api/v1/hermes-4-405b.

curl -X POST https://api.muapi.ai/api/v1/hermes-4-405b \
  -H "x-api-key: $MUAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{"prompt": "Explain how attention works in transformers."}'
  1. Poll or stream: Poll GET /api/v1/predictions/{request_id}/result until status is completed, or stream tokens live from /api/v1/hermes-4-405b/stream.
curl -X POST https://api.muapi.ai/api/v1/hermes-4-405b/stream \
  -H "x-api-key: $MUAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{"prompt": "Explain how attention works in transformers."}'
  1. Read the result: The generated text is returned in outputs. Usage is billed on actual token counts.
❓

Common Questions

Frequently asked

What is Hermes 4 405B?

Hermes 4 405B is a 405B hybrid-reasoning chat model from Nous Research with a 128K-token context window, known for steerability and low refusal rates.

How does it handle refusals?

Hermes 4 405B is tuned to refuse less often than mainstream assistants, but it does not guarantee unrestricted output and can still decline some requests.

How is Hermes 4 405B priced?

Billing is per token: $0.6/M input tokens and $2.4/M output tokens. A small minimum is reserved when the request starts and reconciled to actual usage.

Does Hermes 4 405B support streaming?

Yes. Stream tokens live from `/api/v1/hermes-4-405b/stream` using Server-Sent Events, or poll the async endpoint.

What is the context window?

The context window is about 128K tokens. Prompt and output together must fit inside it; `max_tokens` is capped at 32768. The endpoint accepts text input only.

Is output guaranteed to be unrestricted?

No. Behavior varies by prompt, and you remain responsible for complying with applicable laws and Muapi's acceptable-use policy.