Nousresearch
Hermes 4 405B
Released 2025-08-26 · Knowledge up to 2024-08-31
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...
You send / 1M
$1.00
tokens in
It replies / 1M
$3.00
tokens out
Memory
131K
tokens it can read at once
Typical request
$0.0028
1,000 words in, 400 out
Data privacy
No training or retention: Yes. At least one provider serves it with zero data retention.
| Provider | You send / 1M | It replies / 1M | Uptime | Privacy policy |
|---|---|---|---|---|
| Nebius · fp8 | $1.00 | $3.00 | 100% | Read |
Based on OpenRouter and Vercel AI Gateway data. Confirm terms with the provider before sending sensitive data.
What it can do
- Reads: text
- Writes: text
- Reasoning: Yes
- Tool calling: No
- Longest reply: 118K tokens
- Repeated text is billed at $0.200 per 1M (prompt caching)
- Intelligence score: 7.5 / 100 (Artificial Analysis)
Similar models
- GLM 5Z.ai · $1.00 in / $3.20 out per 1M
- Qwen3 235B A22B Thinking 2507Qwen · $0.450 in / $3.50 out per 1M
- Hy4 previewTencent · $0.834 in / $2.50 out per 1M
- Kimi K2 ThinkingMoonshot · $0.600 in / $2.50 out per 1M
- Nemotron 3 UltraNVIDIA · $0.600 in / $2.40 out per 1M
- DeepSeek V4 Pro 0813DeepSeek · $1.32 in / $3.96 out per 1M