Swiss AI releases Apertus v1.5 with 70B and 8B open multimodal models
jacek2023 · reddit · 2026-07-24
Swiss AI released Apertus v1.5, a family of 8B and 70B open language models aimed at multilingual, transparent AI.
- Supports text, images, and audio as input, generating text
- Context length goes up to 262,144 tokens
- The 8B model was continued-pretrained with 4T tokens of multimodal data; the 70B model with 2T tokens
- Uses a decoder-only transformer with xIELU and AdEMAMix
- Adds a new thinking mode for reasoning and improved tool use
- The team says the models were trained on fully open data and will publish the technical report, benchmarks, pipelines, and intermediate checkpoints later
Related event: Swiss AI Releases Apertus 1.5 Open Multimodal Models(3 posts)→
More from Multimodal
- Dense instance segmentation via Muse Wonder + SAM 3.1 on Meta's Model API — nikhilaravi · 2026-10-26
- LichtFeld Studio v0.5.4.3 ships community-built SteadySplats, POPSpa and Voxelization — janusch_patas · 2026-10-26
- One prompt to a 2-3 min narrated short: hands-on test of Antigravity video generation — AI_Andrew · 2026-10-26
- Creator makes 10,564-frame animated short entirely with Gemini toolchain — AI_Andrew · 2026-10-26
- FLUX.3 Klein step-distilled to 4-8 steps with CFG baked in — Fit-Structure-6489 · 2026-10-26
- ostris pre-training "Primal VAE": 1-channel latents match modern 8x VAEs via patching — ostrisai · 2026-10-26