Key Signals

• Claude Code ships two releases addressing enterprise connectivity and developer experience. Version 2.1.23 delivers significant improvements for corporate environments, fixing mTLS and proxy connectivity issues that blocked users behind corporate proxies or using client certificates. The release also addresses shared system permission conflicts and prompt caching race conditions, making Claude Code more reliable in enterprise deployment scenarios. [1][2]

• OpenAI reveals internal use of agentic AI for data analysis at scale. The company disclosed how they built an in-house data agent combining GPT-5, Codex, and memory systems to reason over massive datasets. This signals OpenAI's confidence in their own agentic coding tools for mission-critical internal workflows, and provides a reference architecture for enterprises building similar data-driven AI agents. [3]

• OpenAI accelerates model deprecation timeline, retiring multiple ChatGPT models. GPT-4o, GPT-4.1, GPT-4.1 mini, and o4-mini will be retired from ChatGPT on February 13, 2026, alongside the previously announced GPT-5 variant retirements. Notably, API access remains unchanged, suggesting OpenAI is consolidating the consumer product around newer model generations while maintaining backward compatibility for developers. [4]

• Claude Code improves terminal UX with customizable spinners and better timeout visibility. The v2.1.23 release introduces a spinnerVerbs setting for personalized feedback during operations, and now displays timeout duration alongside elapsed time for Bash commands. These quality-of-life improvements reduce developer frustration during long-running operations and improve debugging workflows. [1]

Feature Update

• Claude Code v2.1.23 delivers a substantial update focused on enterprise reliability and developer experience. The release fixes critical mTLS and proxy connectivity issues affecting users behind corporate proxies or using client certificates—a common pain point in enterprise environments. Additionally, per-user temp directory isolation now prevents permission conflicts on shared systems, making Claude Code more suitable for multi-tenant deployments. A race condition causing 400 errors when prompt caching scope was enabled has been resolved, along with fixes for async hook cancellation in headless streaming sessions. Terminal performance receives a boost through optimized screen data layout, and ripgrep search now properly reports timeout errors instead of silently returning empty results. The update also adds visual polish: merged pull requests now display a purple status indicator, and a new spinnerVerbs setting allows customization of loading feedback. [1]

• Claude Code v2.1.25 is a targeted hotfix for gateway users on AWS Bedrock and Google Vertex platforms. The release addresses a beta header validation error that affected these cloud provider integrations. Users encountering this issue can set the environment variable CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS=1 as a workaround. This quick turnaround fix demonstrates Anthropic's commitment to supporting Claude Code across multiple cloud deployment models. [2]

AI Coding News

• OpenAI reveals details about their internal AI data agent combining GPT-5, Codex, and persistent memory for large-scale data analysis. This represents a significant proof-of-concept for agentic AI workflows, demonstrating how reasoning models can be integrated with code execution capabilities and memory systems to handle complex, multi-step data analysis tasks. The disclosure provides valuable architectural insights for organizations building their own AI-powered data pipelines and suggests OpenAI views internal dogfooding as an important validation method for their agentic tools. [3]

• OpenAI announces retirement of GPT-4o, GPT-4.1, GPT-4.1 mini, and o4-mini from ChatGPT on February 13, 2026. This retirement coincides with the previously announced deprecation of GPT-5 Instant, Thinking, and Pro variants. Importantly, the API remains unaffected by these changes, meaning developers building on OpenAI's platform can continue using these models. The aggressive consolidation of consumer-facing models suggests OpenAI is streamlining ChatGPT around its newest model generations while maintaining stability for enterprise API customers. [4]