Model reference · open weights
Hermes-Trismegistus-Mistral is an open-weight language model from teknium. Hermes-Trismegistus-Mistral-7B (BF16) weighs 14.5 GB; the smallest configuration that runs it is 2× RTX 3060 12 GB.
Summary of the teknium/Hermes-Trismegistus-Mistral-7B model card, 2026-10-01
What it is
| Released by | teknium |
|---|---|
| Released | 2023-11-02 |
| VRAM | 14.5 GB for the weights |
What it runs on
| Card | Requests at once | Context max | Memory | |
|---|---|---|---|---|
| 8K each | 32K each | |||
| RTX 3060 12 GB | — | — | — | 11.6 GB |
| RTX 4060 Ti 16 GB | — | — | — | 15.4 GB |
| RTX 3090 24 GB | 7 | 7 | all 32K | 23.4 GB |
| RTX 4090 24 GB | 7 | 7 | all 32K | 23.4 GB |
| RTX 5090 32 GB | 14 | 14 | all 32K | 31.0 GB |
| L40S 48 GB | 26 | 26 | all 32K | 44.0 GB |
| A100 80 GB | 58 | 58 | all 32K | 78.2 GB |
| H100 80 GB | 55 | 22 | all 32K | 78.1 GB |
| RTX PRO 6000 Blackwell 96 GB | 70 | 28 | all 32K | 93.8 GB |
| DGX Spark (GB10) 128 GB unified | 83 | 33 | all 32K | 107 GB |
| H200 141 GB | 111 | 44 | all 32K | 138 GB |
| B200 180 GB | 146 | 36 | all 32K | 176 GB |
| 2× RTX 3060 12 GB tensor parallel | 6 | 6 | all 32K | 11.6 GB a card |
| 2× RTX 4060 Ti 16 GB tensor parallel | 14 | 14 | all 32K | 15.4 GB a card |
| 2× RTX 4090 24 GB tensor parallel | 28 | 28 | all 32K | 23.4 GB a card |
| 2× RTX 3090 24 GB tensor parallel | 28 | 28 | all 32K | 23.4 GB a card |
| Requests at once | 8K tokens each | 32K tokens each |
|---|---|---|
| 1 | 16.2 GB | 16.2 GB |
| 5 | 20.5 GB | 20.5 GB |
| 8 | 23.7 GB | 23.7 GB |
| 16 | 32.3 GB | 32.3 GB |
| 32 | 49.5 GB | 49.5 GB |
| 64 | 83.8 GB | 83.8 GB |
One card, with vLLM's small-card settings.
From the model card
Transcendence is All You Need! Mistral Trismegistus is a model made for people interested in the esoteric, occult, and spiritual.
The change between Mistral-Trismegistus and Hermes-Trismegistus is that this version trained over hermes 2.5 instead of the base mistral model, this means it is full of task capabilities that it Trismegistus can utilize for all esoteric and occult tasks, and performs them far better than ever before.
Here are some outputs:
Special thanks to @a16z.
This model was trained on a 100% synthetic, gpt-4 generated dataset, about ~10,000 examples, on a wide and diverse set of both tasks and knowledge about the esoteric, occult, and spiritual.
The dataset will be released soon!
Prompt Format:
USER:
ASSISTANT:
OR
USER:
ASSISTANT:
No benchmark can capture the nature and essense of the quality of spirituality and esoteric knowledge and tasks. You will have to try testing it yourself!
Training run on wandb here: https://wandb.ai/teknium1/occult-expert-mistral-7b/runs/coccult-expert-mistral-6/overview
Apache 2.0
Quoted from the model card on Hugging Face. The full card is behind the Hugging Face link above.