MythoMax L2 13B

by Gryphe

Mythomax L2 13B is a 13-billion parameter open-weight model released by Gryphe in August 2023, built on Meta's Llama 2 foundation. It was designed to excel at creative writing and roleplay scenarios, combining emotional coherence with character consistency across long narrative exchanges. The model merges MythoLogic-L2 and Huginn using an experimental BlockMerge_Gradient tensor merging technique, where each of its 363 tensors received a unique blending ratio. This approach fuses one model's robust language understanding with another's narrative writing capability to produce strong results in both storytelling and character consistency. Mythomax L2 keeps persona traits, speech patterns, and emotional nuance steady across thousands of tokens without drifting into generic assistant-speak. It resists breaking character in complex roleplay scenarios and sustains emotional coherence in dialogue-driven narratives. The model uses Alpaca prompt formatting, and community quantizations in GGUF, GPTQ, and AWQ formats provide flexibility across varied hardware constraints.

Key info

Input
Output
Features
Context window
4K
Max output
4K

Available routes

No routes currently available — MythoMax L2 13B isn't routed through the Opper gateway right now. It may return.

Contact us about this model →

Start building with 700+ models

One API key for every major provider, up and running in minutes.

Get startedView Documentation