Agent Tools Today, 19 Nov 2026: GPT-6.2 Astra and Sol, Gemini 4 Neon and MAI Code 2 OpenAI released GPT-6.2 Astra and Sol on 17 Nov 2026, Google launched Gemini 4 Neon on 18 Nov, and Microsoft released MAI Code 2 and MAI Code 2 Flash on 17 Nov, according to the companies' own model pages and posts. OpenAI lists a 1,050,000-token context window for both GPT-6.2 models, with Astra priced at $8 per million input tokens and $40 per million output and Sol at $2 and $10, and input and cache rates doubling above 272K input tokens. Google priced Gemini 4 Neon at $3 per million input tokens and $15 per million output tokens with a 1 million-token context window and cited its own figure of 58.6, ranking Neon #1 overall on the Artificial Analysis Coding Agent Index, while Microsoft said MAI Code 2 delivers "98% of the performance of GPT-6.1 Sol at 40% of the cost" at its 50% promotional price and is not yet tuned for very long-horizon agentic work. Agent Tools Today, 19 Nov 2026: GPT-6.2 Astra and Sol, Gemini 4 Neon and MAI Code 2 The top launches, skills, multi agent moves, rising repos and talking points in AI agents from 16 to 19 Nov 2026, each with its source and date. Agent tools today, 19 Nov 2026, in one line: OpenAI shipped GPT-6.2 Astra and Sol, Google launched Gemini 4 Neon, and Microsoft released MAI Code 2. It covers the three days since the 16 Nov brief no brief shipped 17 or 18 Nov , 16 Nov 04:00 UTC to 19 Nov 04:00 UTC, checked 19 Nov. Previous edition: 16 Nov https://munderdiffl.in/blog/agent-tools-today-2026-11-16/ . Top 5 launches top-5-launches 1. GPT-6.2 Astra and Sol. OpenAI’s model pages call Astra “Our most capable model for your hardest projects”. Both list a 1,050,000 token context window. Astra costs $8 per million input tokens and $40 per million output; Sol costs $2 and $10, checked 19 Nov. Above 272K input tokens, “input and cache rates double”. 1,337 points on Hacker News https://news.ycombinator.com/item?id=50535230 , posted 17 Nov at 17:46 UTC. More in GPT-6.2 Astra and Sol https://munderdiffl.in/blog/gpt-6-2-astra-and-sol/ . Source https://developers.openai.com/api/docs/models/gpt-6.2-astra 2. Gemini 4 Neon. Google’s post, dated 18 Nov, prices it at “$3 per million input tokens” and “$15 per million output tokens, with a 1 million-token context window”. It ranks Neon “ 1 overall on the Artificial Analysis Coding Agent Index 58.6 ”: Google’s own figures. 540 points on Hacker News https://news.ycombinator.com/item?id=50552424 , posted 18 Nov at 21:34 UTC. More in Gemini 4 Neon https://munderdiffl.in/blog/gemini-4-neon/ . Source https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-neon/ 3. MAI Code 2 and MAI Code 2 Flash. Microsoft AI’s post, dated 17 Nov, says MAI Code 2 gives “98% of the performance of GPT-6.1 Sol at 40% of the cost”, with cost “shown at its 50% promotional price”: Microsoft’s own figures. Both are “explicitly not yet tuned” for “very long-horizon agentic work”, and both are in GitHub Copilot. They are two of seven new models https://microsoft.ai/news/microsoft-ignite-2026/ Microsoft released for Ignite. 417 points on Hacker News https://news.ycombinator.com/item?id=50535885 , posted 17 Nov at 18:25 UTC. Source https://microsoft.ai/news/introducing-mai-code-2-and-mai-code-2-flash/ 4. Claude Code v2.1.326 to v2.1.328. Three releases, from 16 Nov at 22:11 UTC to 19 Nov at 01:14 UTC. v2.1.327 https://github.com/anthropics/claude-code/releases/tag/v2.1.327 adds claude mcp add-json