Carl Franzen

Alibaba's Qwen Team released Qwen3-Max-Preview, a 1 trillion parameter LLM with competitive performance on benchmarks, available via API and select platforms but not as open source.

“Author of "Qwen3-Max arrives in preview with 1 trillion parameters, blazing fast response speed, and API availa"”

Liquid AI released LFM2-VL, a vision-language model that delivers up to 2x faster GPU inference speed than comparable models while maintaining competitive performance on benchmarks, enabling efficient on-device AI for smartphones and embedded systems.

“Author of "Liquid AI wants to give smartphones small, fast AI that can see with new LFM2-VL model" in VentureBeat”

MiniMax released open-source M2.5 and M2.5 Lightning models that match state-of-the-art performance from Google and Anthropic while costing 95% less, particularly for enterprise agentic tasks.

“Author of "MiniMax's New Open M2.5 and M2.5 Lightning Near State-of-the-art While Costing 1/20th of Claude Opus" in VentureBeat”

“Wrote article analyzing Google's Always On Memory Agent open-source release and its implications for enterprise AI.”

“Reporter covering Alibaba's Qwen3.5 Small Model Series release and AI efficiency benchmarks.”

“Author reporting on OpenAI's Codex app reaching 1M downloads and rate limit changes.”

“Demonstrated Alibaba's Qwen Deep Research tool capabilities in creating reports, webpages, and podcasts.”