Hermes 4 - Llama-3.1 70B
Open weights
Nous Research's mid-size hybrid-reasoning post-training of Llama-3.1-70B with switchable <think> mode and JSON-schema-faithful outputs.
- Publisher
- Nous Research
- Family
- Hermes
- Params
- 70B
- Context
- 131K
- Input
- Output
- License
- Llama 3 Community
- Released
- 27 Aug 2025
Cite
Notes
Only stored in your browser.
Context
131K
Input
$0.13
/1M tok
Output
$0.40
/1M tok
Blended
$0.20
#240/ 461/1M · 3:1
Speed
94
#64/ 461tok/s
Price & speed via Artificial Analysis · ranked across tracked models
| Reasoning effort | Index | Speed |
|---|---|---|
| Reasoning | 9.9 | 92 t/s |
| Non-reasoning | 6.7 | 94 t/s |
Intelligence Index per reasoning-effort setting, via Artificial Analysis · pricing identical across tiers
Reported on 9 evals across 6 domains - best 66.4% on MMLU-Pro · #179 of 340