Happy Horse 1.0 is now on ModelsLab

Try Now
Skip to main content
Available now on ModelsLab · Language Model

Nous: Hermes 4 70BReason Hybrid, Respond Precise

Try Nous: Hermes 4 70BAPI Documentation

Toggle Reasoning, Master Tasks

Hybrid Mode

Control Think Traces

Enable <think> tags for step-by-step reasoning or direct answers with reasoning boolean.

Schema Outputs

JSON Function Calling

Produces valid JSON schemas, supports tool use and function calling for API integrations.

STEM Boost

Math Code Logic

Excels in mathematics, coding, STEM, and logic via 60B token post-training corpus.

Examples

See what Nous: Hermes 4 70B can create

Copy any prompt below and try it yourself in the playground.

Math Proof

“Prove Fermat's Last Theorem for n=3 using step-by-step reasoning in <think> tags, then summarize the key steps in plain English.”

Code Debug

“Analyze this Python function for bugs: def factorial(n): if n == 0: return 1 else: return n * factorial(n-1). Fix and optimize with reasoning enabled.”

JSON Schema

“Generate a weather API response in strict JSON schema: {location: string, temp: number, forecast: array}. Use Paris, France with reasoning for data logic.”

Logic Puzzle

“Solve Einstein's riddle: five houses, colors, nationalities, drinks, smokes, pets. Who owns the fish? Output structured steps then final answer.”

For Developers

A few lines of code.
Reasoning LLM. One Call.

ModelsLab handles the infrastructure: fast inference, auto-scaling, and a developer-friendly API. No GPU management needed.

  • Serverless: scales to zero, scales to millions
  • Pay per token, no minimums
  • Python and JavaScript SDKs, plus REST API
API Documentation 
import requests


response = requests.post(
    "https://modelslab.com/api/v7/llm/chat/completions",
    json={
  "key": "YOUR_API_KEY",
  "prompt": "",
  "model_id": ""
}
)
print(response.json())

FAQ

Common questions about Nous: Hermes 4 70B

Read the docs

Nous: Hermes 4 70B is a 70B parameter LLM from Nous Research on Llama-3.1-70B base. It features hybrid reasoning with 131k context. Released August 2025.

Access via LLM endpoint with reasoning boolean to toggle <think> traces. Supports JSON mode, schema adherence, function calling. Pricing starts at $0.13/M input tokens.

Improves math, coding, STEM, logic over Hermes 3 via 60B token corpus. Offers steerability, reduced refusals, tool use. Maintains general tasks like writing.

Yes, fully open-weight for accessibility. Trained with synthetic data pipeline, no pre-training changes. Benchmarks high on math and refusal tasks.

Alternative to closed models like GPT-4o via open API access. Matches frontier reasoning at lower cost. Use for structured outputs and hybrid mode.

Supports 131,072 tokens. Handles long inputs for complex reasoning chains. Toggle reasoning for speed or depth control.

Ready to create?

Start generating with Nous: Hermes 4 70B on ModelsLab.

Try Nous: Hermes 4 70BAPI Documentation