Blog

Poolside unveils Laguna-XS.2, a 33B MoE coding model for local deployment and agentic workflows.

DeepSeek unveils V4-Pro and V4-Flash, challenging US dominance with massive context and aggressive pricing.

OpenAI releases GPT-5.5 on April 23, 2026, marking a historic milestone in AI with unmatched benchmark scores and token efficiency.

Xiaomi releases MiMo-V2.5-Pro, a 1-trillion parameter MoE model challenging GPT-5 and Claude Opus with native multimodal capabilities and affordable API pricing.

Alibaba releases Qwen3.6-27B, an open-source 27B model outperforming 397B MoE on SWE-bench. Apache 2.0 licensed, multimodal, available on Hugging Face.

An investigation into how Anthropic's Claude Code injects hidden system reminders that consume up to 50% of your context window, costing users millions of tokens without their knowledge.

Moonshot AI has released the Kimi K2.6 model, a milestone open-weights release featuring 262k context and autonomous agent capabilities.

Anthropic releases Claude Opus 4.7, a milestone model delivering unprecedented reasoning capabilities, high-resolution vision, and optimized agentic workflows for developers.

Zhipu AI releases GLM-5.1, a 744B MoE model trained on Huawei Ascend chips that dominates SWE-Bench Pro and offers MIT licensing for developers.