Yandex releases open AliceAI-Foundation-80B-A3B-Base, aimed at Qwen and DeepSeek Flash
Iwaku_Real · reddit · 2026-09-22
Yandex has released the open-weights base model AliceAI-Foundation-80B-A3B-Base on Hugging Face — a sparse MoE with 80B total / 3B active parameters, positioned as a Russian competitor to Qwen 35B-class and DeepSeek V4 Flash. Caveat: no llama.cpp support yet, so local deployment options remain limited.
Related event: Yandex Open-Sources AliceAI 80B MoE Model with 256K Context(3 posts)→
More from Models
- YC-backed Monte launches post-training platform to build proprietary models from your agents' real work — ycombinator · 2026-11-12
- Leak Claims OpenAI Will Launch Entire GPT-6.2 Family Tomorrow to Counter Opus 5.5 and Fable 5.2 — heypearlai · 2026-11-12
- ARC-AGI-3 leaderboard: models score 90%+ but at $10K+ per task — haider1 · 2026-11-12
- LMArena Launches Agent Arena for Head-to-Head Agent Evaluation — arena · 2026-11-12
- LLMs lack hyperlocal municipal data, explains disputed Texas power outage debate — csuwildcat · 2026-11-12
- Claude API's per-message effort lets you switch reasoning levels without breaking prompt cache — lyraxana · 2026-11-12