Olala v1.0 Mini 7A1B
Small reasoning model, 6.8B total / ~1B active: chat, tool calling, translation, summarization. Open data, Apache 2.0.
Text Generation • 7B • Updated • 1.02k • 177Note Start here: chat, reasoning and tool calling, with ~1B active parameters.
OVHaiLLM/Olala-v1.0-mini-7A1B-GGUF
Text Generation • 7B • Updated • 7.09k • 73Note Laptops, CPUs and NVIDIA GPUs with the llama.cpp fork. About 70 tokens/s on a 16 GB MacBook Air M5.
OVHaiLLM/Olala-v1.0-mini-7A1B-MLX
Text Generation • Updated • 5Note Apple Silicon Macs with the mlx-engine fork. Faster than llama.cpp on a Mac.
OVHaiLLM/Olala-v1.0-translate
Text Generation • 7B • Updated • 69 • 4Note Machine translation in 18 languages: follows your terminology and keeps casing.
OVHaiLLM/Olala-v1.0-summarise
Text Generation • 7B • Updated • 68 • 3Note Document summarization: structured, faithful summaries.
OVHaiLLM/Olala-v1.0-mini-7A1B-base
Text Generation • 7B • Updated • 99 • 7Note Pre-trained base model (12.8T tokens of public and own synthetic data), not aligned: for fine-tuning and research only.
OVHaiLLM/Olala-v1.0-mini-7A1B-SFT
Text Generation • 7B • Updated • 80 • 7Note After supervised fine-tuning, before alignment (not aligned): a start for your own RL or fine-tuning.