Insights &amp; News

Anthropic offers Claude credits for rare-disease AI for Science research

Anthropic announced a rare-genetic-disease research call on July 20, offering accepted applicants up to $50,000 in Claude credits for six months.

Jul 22, 2026

Gemini 3.5 Flash Cyber brings a lightweight model into CodeMender’s defense loop

Google DeepMind introduced Gemini 3.5 Flash Cyber on July 21, describing a lightweight model that CodeMender can call repeatedly to find, validate, and patch vulnerabilities.

Jul 22, 2026

OpenAI Presence puts enterprise agents into customer and internal workflows

OpenAI introduced Presence on July 22 as a deployment product for voice and chat agents that can use approved systems, follow policies, and escalate to people.

Jul 22, 2026

GitHub Code Quality reaches GA as AI-assisted coding adds a governance layer

GitHub Code Quality is generally available on Enterprise Cloud and Team, combining deterministic CodeQL analysis, AI-assisted detection, coverage metrics, quality gates, and Copilot Autofix.

Jul 21, 2026

Gemini 3.6 Flash enters GitHub Copilot as model choice moves closer to long-running agents

GitHub is rolling out Gemini 3.6 Flash to Copilot with configurable reasoning and parallel tool use for coding, web development, and longer-horizon agent workflows.

Jul 21, 2026

Microsoft brings AMD AI infrastructure to Azure as agent coordination becomes a CPU concern

Microsoft’s July 20 Azure update brings AMD Helios and next-generation EPYC systems to HDv2, HXv2, and ND MI455X v7 for data systems, chip design, inference, and agent coordination.

Jul 21, 2026

Kimi K3 brings a 2.8T open frontier model to long-horizon coding agents

Moonshot AI introduced Kimi K3 on July 16 with 2.8T parameters, native vision, a 1M-token context window, and long-horizon coding workflows; full weights are due July 27.

Jul 20, 2026

Alibaba Cloud’s WAIC 2026 update moves from model platforms to an agent runtime layer

Alibaba Cloud’s July 20 WAIC update combines AgentRun, AgentLoop, AgentTeams, TokenWorks, Qwen 3.8-Max-Preview, and T-Head SAIL into an agent-native platform story.

Jul 20, 2026

NVIDIA at SIGGRAPH: MCP connects creative tools to agents as physical AI moves local

NVIDIA’s July 20 SIGGRAPH update connects creative applications through MCP and pairs local agents with synthetic-video detection, edge world models, and governed runtimes.

Jul 20, 2026

Qwen-Coder-Qoder puts model, agent, and product in one coding flywheel

Alibaba Cloud Community introduced Qwen-Coder-Qoder on July 17, 2026, describing a Qwen-Coder model aligned with Qoder's tools, context, and agent architecture.

GitHub Copilot code review adds custom instructions, setup steps, and firewall controls

GitHub's July 17, 2026 Copilot code review update adds branch-specific instructions, custom setup steps, a default firewall, and independent runner settings.

Grok Build open-sources the harness behind its coding agent

SpaceXAI open-sourced Grok Build on July 15, 2026, exposing its agent loop, tools, terminal UI, and skills, plugins, MCP, and subagent extension system.

Grok Automations turns one-off prompts into scheduled and email-triggered jobs

SpaceXAI introduced Grok Automations on July 16, 2026, letting users create jobs that run on schedules or when matching emails arrive.

Google DeepMind frames bioresilience as both AI safety and outbreak response

Google DeepMind and Isomorphic Labs outlined a bioresilience approach on July 16, 2026, covering prevention, detection, response, and trusted AI partnerships.

GitHub Copilot usage metrics now expose repository-level agent activity

GitHub made repository-level Copilot usage metrics generally available on July 17, 2026, covering pull-request activity from coding agents and code review.

NVIDIA Vera Rubin frames agentic post-training around intelligence per dollar

NVIDIA's July 17, 2026 post describes continuous post-training for agentic models and introduces intelligence per dollar as a complement to cost per token.

Grok 4.5 brings coding, agentic tasks, and office work into one model

SpaceXAI launched Grok 4.5 on July 16, 2026, positioning it for coding, agentic tasks, and knowledge work across Grok Build, Cursor, and the API.

GitHub expands secret scanning with AI-detected secret categories and push protection

GitHub's July 15, 2026 update adds Resend and APIclub detectors, default VolcEngine push protection, and webhook categories that identify AI-detected secrets.

Jul 17, 2026

Anthropic’s Claude for Teachers connects curriculum standards to scheduled agent workflows

Anthropic's July 14, 2026 Claude for Teachers gives verified US K-12 educators free premium access, teaching skills, curriculum connections, and scheduled workflows.

Jul 17, 2026