TAAFT
Free mode
100% free
Freemium
Free Trial
Deals
Companies
Swiss AI Initiative
Models
Apertus 8B Instruct 2509

Apertus 8B Instruct 2509

By Swiss AI Initiative
Model family: LLaMA
Apertus 8B Instruct 2509 is the instruction-tuned variant of the Apertus 8B base model, a decoder-only transformer pretrained on 15T tokens using a staged curriculum of web, code, and curated data. It natively supports 1,811 languages, positioning it among the most massively multilingual open models available. The model is fully open: weights, training data, and training recipes are all publicly released. It is built with compliance in mind, respecting data owner opt-outs and avoiding sources with restrictive terms of service. The context window extends to 65,536 tokens. The model supports tool use for agentic workflows. It was trained on 4,096 GH200 GPUs using Megatron-LM and is deployable via Transformers, vLLM, SGLang, and MLX. Performance on general language understanding benchmarks is competitive with Llama 3.1 8B. Licensed under Apache 2.0.
Text Gen 7
Released: September 17, 2025

Overview

Apertus 8B Instruct 2509 is an 8B parameter open-source instruction-tuned language model pretrained on 15T tokens. It natively supports 1,811 languages, uses fully open data and weights with disclosed training recipes, and is designed to respect data owner opt-outs. It features a 65,536-token context window and supports tool use for agentic tasks.

💬Chatting🌐Text translation💻Coding📝Writing🔍Research

About Swiss AI Initiative

The Swiss AI Initiative is the world's largest open science/open source effort for AI foundation models, started in December 2023. Seeded with over 10M GPU hours on the Alps supercomputer and a 20M CHF grant from the ETH Domain, it is the first initiative of the Swiss National AI Institute—a partnership between the ETH AI Center and the EPFL AI Center—leveraging 800+ researchers (70 AI-focused professors) from 10+ Swiss academic institutions.

Industry: Research
Location: Zürich, CH
Website: swiss-ai.org
View Company Profile

Other models from this family

View all models from this family

Related Models

  • Gen 7
    Gemma 2
    💻Coding
    Text
    Released 2y ago
  • Gen 7
    Harrier OSS v1 0.6b
    Harrier OSS v1 0.6B is Microsoft’s multilingual text embedding model for semantic search, retrieval, clustering, similarity, classification, bitext mining, and reranking. It uses a decoder-only architecture with last-token pooling and L2 normalization, supports 94 languages, handles up to 32,768 tokens, and is positioned as a strong multilingual embedding model.
    Text
    Released 3mo ago
  • Gen 7
    Llama 3.3
    Llama 3.3 70B is Meta’s updated 70B model in the Llama 3.x family, tuned for stronger reasoning, coding, and instruction following. It supports long-context prompting, tool/function calling, and reliable JSON outputs, with open weights under the Llama license.
    💻Coding📝Writing🚀Productivity🙋Text humanization
    Text
    Released 1y ago
Last updated: June 30, 2026
0 AIs selected
Clear selection
#
Name
Task