Nous: Hermes 3 70B Instruct
Nous: Hermes 3 70B Instruct is designed for handling text inputs within a context length of 131,072 tokens. While it excels in processing textual data, it does not support reasoning, structured output, or integration with external tools. This model might be worth considering if you are looking to manage large volumes of text-based tasks without the need for complex reasoning or tool interactions. Its price is competitive at $0.7 per million input and output tokens, making it suitable for projects requiring significant text processing, though its benchmark score of 6.4 across one independent benchmark suggests there may be other models with stronger performance in specific areas.
Benchmark results
Independent, published benchmarks. Blended score 6.4 across 1 benchmark, last refreshed 2026-08-13. How scoring works →
| Benchmark | Measures | Score |
|---|---|---|
| AI Index | broad capability composite | 7.9 |
- Model ID
- nousresearch/hermes-3-llama-3.1-70b
- Vendor
- nousresearch
- Released
- August 2024
- Tokenizer
- Llama3
- Input Modalities
- text
- Output Modalities
- text
- Max Output
- 16,384 tokens
- Tool Calling
- not supported
- Structured Output
- ✓ supported
- Reasoning Mode
- not supported
- Vision
- text only
- Audio
- no
- Moderated
- no
What it costs in practice
Computed from the current $0.70/M input and $0.70/M output rates. Run your own numbers →
| Job | Tokens | Cost |
|---|---|---|
| Summarize a 50-page report | 30k in / 1.5k out | $0.02 |
| Classify 1,000 customer emails | 500k in / 50k out | $0.38 |
| A month of a busy support chatbot | 5M in / 2M out | $4.90 |
Similar models
Quick answers
- How much does Nous: Hermes 3 70B Instruct cost?
- $0.70 per million input tokens and $0.70 per million output tokens.
- What is Nous: Hermes 3 70B Instruct's context window?
- 131,072 tokens, roughly 196 pages of text in a single request.
- Does Nous: Hermes 3 70B Instruct support tool calling?
- It supports structured output.
- Can Nous: Hermes 3 70B Instruct process images?
- No, it is text-only on the input side.