Product

Llama

2 articles First covered Apr 27, 2026, latest Aug 10, 2026

AMD's glass-fronted Santa Clara headquarters behind ornamental grasses, with the AMD sign on a low monument by the sidewalk

One model per chip: AMD bought Taalas, whose test part hit 17,000 tokens a second

Taalas casts a model's weights into transistors. Its 6nm test chip served Llama 3.1 8B at roughly 17,000 tokens per second, and AMD just bought the company.

Arcee AI Trinity branding from the Trinity-Large-Thinking blog post.

Open Source·3 months ago

Arcee's Trinity-Large-Thinking is a 399B open MoE that costs 96% less than Opus

Arcee released Trinity-Large-Thinking on April 1: a 399B-param sparse MoE with 13B active, Apache 2.0 weights, $0.88 per million output tokens, and PinchBench just behind Opus 4.6.