The AI Coding Supremacy wars
Vibe working, Enterprise AI Coworkers, the Frontier Keeps evolving. Capex pushes the Enterprise AI landscape into more real-world trials as the SaaSpocalypse fears collide.
Good Morning,
It’s been a hectic week in the AI News cycle especially around the future of coding.
OpenAI dropped GPT-5.3-Codex
Anthropic released Claude Opus 4.6
Alibaba Cloud team released Qwen3-Coder-Next
OpenAI also released OpenAI Frontier.
This as we are anticipating DeepSeek’s own coding model referred to as DeepSeek V4 in about 10 days time. Meanwhile before we get Grok 5, we are likely to get Gemini 3.5 (codenamed "Snow Bunny") has been spotted in Google’s internal sandboxes (LM Marina). Then DeepSeek’s flagship DeepSeek-R2 model later in March next month.
What’s Next
GLM-5 (Zhipu AI)
DeepSeek V4
Kimi K2.5’s “Agent Swarm” API
Gemini 3.5
Grok 5
Claude Sonnet 5
DeepSeek-R2
Qwen 3.5 & "Max-Thinking"
Llama 4 Behemoth (April)
As adoption of Claude Code accelerates, it’s getting harder to compare the capabilities of Anthropic and OpenAI’s coding tools (but we are going to try).
Claude Code Github Commits Skyrockets before SaaSpocalypse takes shape
“4% of GitHub public commits are being authored by Claude Code right now.
At the current trajectory, we believe that Claude Code will be 20%+ of all daily commits by the end of 2026.” - Dylan Patel, SemiAnalysis
Unfortunately, Anthropic didn’t share SWE Bench Pro benchmarks but Sebastian Raschka, PhD of Ahead of AI Newsletter put them side by side based on the available Terminus 2.0 numbers:
I asked Jeff Morhous for details on how OpenAI’s Codex works for some of our more technically inclined readers.
Jeff writes:
The AI-Augmented Engineer
Lately Jeff has been breaking down guides as a Software Engineer that can help us understand the frontier of AI coding tools better:
Claude Code vs Cursor
How to use AI to write code without skill atrophy
How to get unique designs out of Claude Code
Claude Code is the Inflection Point
To understand how the Codex agent loop works, this is a great resource. This article will go into some detail about how Codex actually works. I will also talk about Opus 4.6 vs. GPT-5.3-Codex, the SaaSpolapyse and other relevant topics like Capex. Financial services and banking employees could be most vulnerable to the vibe-working era. The micro-hysteria that AI coding generalized to other industries will cause is now officially a thing.
AI moving toward a ‘vibe working’ era
 | A guest post by
|