Open-source AI has surpassed closed models in token throughput and matched them in coding ability, but production deployment still lags due to operational issues.
PrismML has released Bonsai 27B, a model that compresses Qwen 3.6-27B down to 3.9GB via 1-bit quantization.
Capital One has open-sourced VulnHunter, an agentic AI security tool built on Claude Opus 4.8.
AI inference cloud startup General Compute secured a $400 million loan backed by inference-only chips.
zerostack, a Rust-based coding agent that limits memory usage to around 16MB, has been released, making it over 20 times lighter than JavaScript runtime-based tools.
The Wan Team has released WanSong, a diffusion-based music generation model.
OpenAI has proposed a way to measure AI investment returns using four metrics centered on 'useful intelligence per dollar.'
Axint, an MCP server that automates iOS build verification for AI coding agents, has been released under the Apache 2.0 license.
Vercel Sandbox has stopped charging for downloading external data.
A price trend analysis of Claude models based on Anthropic's official data: the top-tier model has remained stable at $11→$10 over two years, Sonnet's price has stayed fixed, and Haiku has seen a 4x price increase.
The GLM 5.2 model is available at a 35% discount through the Novita route on AI Gateway until July 24.
Ramp has expanded its feature that tracks and manages token spend across multiple AI services like OpenAI, Anthropic, and Google Gemini in a single dashboard.
Google has started supporting connections to major services like Instacart, Canva, and YouTube Music in Search's AI Mode.
Nvidia-backed Fireworks has reached a $17.5 billion valuation, with revenue surging on the back of growing demand for low-cost AI models.
Task-specific routing built by teams with deep expertise in a particular task outperforms general-purpose routers on both cost and accuracy.
GitHub officially launched the Copilot SDK, which supports 6 programming languages.
Anthropic unveiled an LLM reasoning tool called Schema, which achieved 99% performance on the ARC-AGI benchmark.
Anthropic used Claude Code to complete large-scale code migrations that used to take years in just weeks.
The Harness Handbook has been introduced to organize the implicitly scattered behaviors of AI agents into a systematic behavior map, helping with understanding, monitoring, and modification.
Codex's GPT-5.6 offers three models—Sol, Terra, and Luna—that you can choose based on task complexity.
LM Studio has launched Bionic, an AI agent built on open-source models.
Google's Gemini 3.5 Pro launch has been delayed due to insufficient coding capabilities.
Vercel improved cache efficiency and deployment speed by automatically reusing immutable static assets across deployments.
A model called VICIS has been proposed that enables Vision-Language Models to infer visual concepts from example images and apply them.
AI Briefing