All models
Nous Research
Hermes 4 405B frontierruns locally
nous/hermes-4-405b
Hermes 4 is a large-scale reasoning model built on Meta-Llama-3.1-405B and released by Nous Research. It introduces a hybrid reasoning mode, where the model can choose to deliberate internally with...
Input / 1M
$1.00
USD per million tokens
Output / 1M
$3.00
USD per million tokens
Est. cost / token
$0.0000015
$1.50 per 1M, 3:1 mix
Context window
131K
max output 118K
Major features
Open weightsCachingReasoningJSON schema
Task suitability (heuristic 1–5, not a benchmark)
- coding5
- reasoning5
- math5
- summarization5
- extraction5
- classification5
- chat5
- creative5
- translation5
- vision1
- long context3
- agentic3
Run it locally
Will it run on my machine? →| Quantization | Quality | Memory needed (8K ctx) |
|---|---|---|
| Q2_K | poor | 185 GB |
| Q3_K_M | fair | 214 GB |
| Q4_K_Mrecommended | good | 267 GB |
| Q5_K_M | very good | 309 GB |
| Q6_K | excellent | 356 GB |
| Q8_0 | near-lossless | 459 GB |
| F16 | lossless | 859 GB |
Estimates.
Price history
| Recorded | Input | Output | Cached | Source |
|---|---|---|---|---|
| 2026-11-05 03:20 | $1.00 | $3.00 | $0.20 | openrouter |
A row is added whenever a sync detects a price change. Vendor pricing source
Pricing (USD per 1M tokens)
- Input
- $1.00
- Output
- $3.00
- Cached input
- $0.20
- Blended (3:1)
- $1.50
Details
- Company
- Nous Research
- Version
- 4
- Family
- Hermes
- Released
- 2025-08-26
- Parameters
- 405.9B
- Max output
- 118K
- Open weights
- Yes
- Inputs
- text
- Last synced
- 21 min ago