--- title: Anthropic and OpenAI Release Competing AI Coding Models Within Minutes of Each Other description: Anthropic’s Opus 4.6 and OpenAI’s GPT-5.3 Codex launched minutes apart on February 5, revealing contrasting trajectories in the AI coding race. author: Darie Nani (Editor-in-Chief) date: 2026-02-06T16:07:23.000Z updated: 2026-05-05T19:29:40.927Z canonical: https://www.sovereignmagazine.com/article/anthropic-and-openai-release-competing-ai-coding-models-within-minutes-of-each-other image: https://cdn.nanimediahouse.com/claude-46-codex-53-are-out.webp categories: Artificial Intelligence content_type: News region: United States publication: Sovereign Magazine --- Anthropic and OpenAI [released competing agentic coding models](https://www.sovereignmagazine.com/article/openai-unveils-new-api-s-for-gpt-3-5-turbo-and-whisper-models) within 27 minutes of each other on 5 February. Anthropic launched Opus 4.6 at 9.45am Pacific time, moving its announcement forward by 15 minutes from the originally scheduled 10am slot. OpenAI followed with GPT-5.3 Codex shortly after. ## Extending the Lead Anthropic’s Opus has become the industry benchmark. Competing models are routinely judged against its performance across [coding tasks](https://www.sovereignmagazine.com/article/cielara-code-ai-coding-localization). Opus 4.6 aims to extend that lead with agent teams, which allow [multiple Claude instances](https://www.sovereignmagazine.com/article/potpie-ai-raises-2-2-million-to-give-ai-agents-codebase-context) to work in parallel on separate parts of a project. The agents coordinate autonomously without human intervention. Anthropic demonstrated the capability by tasking 16 agents with building a C compiler from scratch. Scott White, Head of Product at Anthropic, said the feature transforms Opus from a tool primarily used by software developers into one suited for product managers, financial analysts, and other knowledge workers. The model scores 65.4% on Terminal-Bench 2.0, an agentic coding benchmark, and 72.7% on OSWorld, which tests computer use tasks. If successful, Opus 4.6 would represent another setback for OpenAI, which already absorbed criticism when ChatGPT-5 launched to underwhelming reception. The model failed to deliver the performance leap users expected and retained flaws from [earlier versions](https://www.sovereignmagazine.com/article/gpt-4-openai-s-latest-breakthrough-in-deep-learning). OpenAI’s GPT-5.3 [Codex](https://www.sovereignmagazine.com/article/openai-codex-plugins-ai-coding-tools) claims 77.3% on Terminal-Bench and positions itself as capable of creating complex games and applications over several days. The company says the model was instrumental in debugging itself during development. GPT-5.3 Codex is 25% faster than its predecessor and is marketed as expanding who can build software. ## From First Mover to Uncertain Future [OpenAI was first out of the gate](https://www.sovereignmagazine.com/article/openai-adoption-technology-s-ethical-challenges-and-limitations). ChatGPT became the generic term for AI chat interactions, the “did you Google that” of the AI age. The company’s enterprise share has collapsed from roughly half the market in 2023 to 27% in 2026, according to Menlo Ventures. Anthropic’s share rose from 18% to 29% in the same period, overtaking OpenAI as the preferred enterprise choice. OpenAI is [projecting $14 billion in annual losses](https://www.sovereignmagazine.com/article/from-microsoft-s-13-billion-ai-investment-to-chatgpt-s-potential-and-risk) by 2026 and does not expect profitability until 2029. Its [ChatGPT Pro subscription](https://www.sovereignmagazine.com/article/openai-s-hidden-25-billion-ad-goldmine-why-790-million-free-users-could-transform-ai-economic), priced at $200 per month, is reportedly losing money. The company has committed $1.4 trillion to data centre construction over eight years while generating $19 billion in annual revenue. Sam Altman called a code red in November after Google released Gemini with performance that reportedly exceeds GPT-5. Several senior researchers have left OpenAI since late 2023, including chief scientist Ilya Sutskever, chief technology officer Mira Murati, and GPT lead author Alec Radford. ## Why This Release Matters More for OpenAI For Anthropic, [Opus 4.6 solidifies its position](https://www.sovereignmagazine.com/article/claude-mythos-anthropic-new-model) as the premier AI coding platform. The company is expanding from developer tools into adjacent enterprise markets while maintaining the technical benchmark others are measured against. Its valuation climbed from $61.5 billion in March 2025 to a reported $350 billion target in early 2026. It now serves more than [300,000 business customers](https://www.sovereignmagazine.com/article/cowork-is-a-big-disappointment-285-billion-wiped-because-of-this). For OpenAI, GPT-5.3 Codex represents something closer to necessity. The company needs to demonstrate that its technology can generate sustainable enterprise income and reverse its market share decline. Without that, the gap between infrastructure commitments and revenue generation becomes untenable. [Three significant Chinese technology companies released their own major AI models](https://www.sovereignmagazine.com/article/the-other-ai-releases-you-missed-this-week) in the same week, indicating the global scale of the AI race. ## Further Context **Q: What are agentic coding models?** Agentic coding models are AI systems that can autonomously write, debug, and modify code across multiple files and sessions. Unlike earlier AI coding assistants that primarily suggested code completions, agentic models can manage entire projects, execute commands, and coordinate complex tasks without continuous human direction. **Q: How do agent teams differ from single AI coding assistants?** Agent teams allow multiple AI instances to work simultaneously on different parts of a project, coordinating their efforts autonomously. A single assistant works sequentially through tasks. Agent teams can divide a large codebase review or multi-component build across parallel agents, reducing total completion time. **Q: Why is enterprise market share important for AI companies?** Enterprise customers provide steadier revenue than consumer subscriptions and typically pay higher rates for guaranteed uptime, security features, and dedicated support. Companies with strong enterprise adoption can forecast income more reliably, which matters when justifying infrastructure investments that run into billions of pounds. **Q: What caused OpenAI’s market share decline?** Multiple factors contributed. GPT-5 underperformed expectations. Chinese company DeepSeek demonstrated competitive models could be built with substantially lower computing requirements. Google’s Gemini reportedly exceeded GPT-5 in performance. Anthropic’s Claude Code gained traction among developers. These developments collectively eroded OpenAI’s early advantage.