AI Models & LLM News: Releases, Benchmarks, Comparisons
New models, benchmarks and our tests
FrontierCode 2.0: Opus 5.5 tops at 45.5%, down from 54.6%
Qwen embedding model searched 20M satellite patches in a weekend
Liquid AI's 2B model hits 96.2% on LIBERO, under 50 ms on Jetson
The new gpt-6.1 terra beat sonnet and kimi – 3.5x cheaper and 3.5x faster than kimi
GPT-6.1 Terra: half the price of Sol, 10 points behind Opus 5.5
AI physics benchmark: GPT-6-Astra solved 49 of 50 unpublished problems
Block's Dessa runs on a MacBook and learned to talk without text
We interviewed Opus 5.5 and DeepSeek and built them as they see it
Kimi K3.1 Pro cuts API prices 33% and leads 6 of 14 Moonshot tests
Claude Mythos 5 kept attacking a real GitHub project in 98% of replays