L3.3 Euryale 70B
Neural Network
L3.3 Euryale 70B is a Sao10k model for roleplay dialogue and creative writing, successor to Euryale L3 70B v2.2.
Max answer length
(in tokens)
Context size
(in tokens)
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works L3.3 Euryale 70B?
L3.3 Euryale 70B is a Sao10k development focused on creative roleplay: the model engages in dialogue as a character, maintains personality and style, and produces vivid creative text instead of dry assistant responses. It continues the line, succeeding Euryale L3 70B v2.2. In BotHub, the model connects without VPN or foreign cards: you pay with a Russian card in rubles only for tokens actually used, and the balance does not expire. Over 250 other neural networks are available in the same interface, so you can compare responses or switch models in a few clicks, or via a unified OpenAI-compatible API by changing one parameter, without rewriting the integration. Chats are encrypted via AES-GCM, and dialogue content is not saved. The model is useful if you write interactive stories and quests, run text-based roleplay campaigns, prototype characters for games or apps, seek a co-author for prose and dialogue, or build a chatbot with a distinct personality.Frequently asked questions about L3.3 Euryale 70B
You can use generated results for commercial purposes. You own all rights to the content you create. The only restriction: make sure your prompt does not include copyrighted third-party material. You are responsible for respecting the rights to any input data.
This is a 70-billion parameter text model from sao10k, fine-tuned based on Llama 3.3. It maintains a given role and style well, making it a choice for creative writing, dialogue, characters, scripts, as well as standard tasks like text rewriting and editing.
The output limit is up to 16,384 tokens per response, which is the volume of a short chapter or a long article. The context window is 8,000 tokens, so very long dialogues should be summarized or provided to the model as a recap of previous parts.
A separate step-by-step reasoning mode is not noted in our data; the model answers immediately. If a chain of thought is needed, ask it in the prompt to break down the task step-by-step. For complex math and logic tasks, it is more convenient to switch to a reasoning model in the same BotHub window.
Support for function calling and strict structured output is not noted in our data. In practice, you can describe the required format directly in the prompt and request a JSON response, but the result should be verified with a validator on your end.
Text, documents, and images are accepted as input — you can send a file with material and ask it to analyze, summarize, or rewrite it. Audio and video are not supported: for those, BotHub has separate models available in the same interface and via a unified API.
The model is built on Llama 3.3 70B, which was originally trained as multilingual, so it works confidently with Russian queries. If the result is not satisfactory, you can compare the response with another model in the same BotHub window without rewriting the integration.