The Battle of Coding Tools: Two Giants and a Dark Horse

The landscape of coding tools is being completely rewritten! On one side, OpenAI decisively shifts its focus from ChatGPT to the Codex platform; on the other, Anthropic experiences explosive growth, doubling its revenue to 204.9 billion RMB, securing its position in the enterprise coding sector with Claude Code.

While many anticipated a showdown between these two giants, an obscure third-party tool, Verdent, quietly emerges as a contender. Many coding professionals may not yet realize that the battle for control over coding tools has shifted, and the ultimate beneficiaries may not be OpenAI or Anthropic.

Key Technologies Explained:

• Codex: OpenAI’s enterprise coding tool, now prioritized over ChatGPT, leverages OpenAI’s model capabilities to focus on enterprise coding services. It is currently not open-sourced and primarily offers paid services to enterprise clients.
• Claude Code: Anthropic’s core coding product, which has driven its revenue surge, is tailored for enterprise transactions. It also has no publicly available open-source information and primarily offers customized paid models.
• Verdent: A third-party multi-model coding tool that integrates mainstream models like Claude, GPT, Gemini, Kimi, and GLM. Its core value lies in its orchestration layer, currently offering a free trial version, with pricing for the paid version yet to be disclosed.

Core Breakdown: Three Players Compete with Unique Strengths

OpenAI’s Shift: Codex Takes Priority Over ChatGPT

According to a deep report by the Financial Times, OpenAI has completed a strategic shift, officially prioritizing Codex over ChatGPT. This change is not a blind trend but a response to market pain points—enterprise coding tools yield higher profit margins than subscription services aimed at general users.

To consolidate its advantage, OpenAI has secured 8 gigawatts of computing power, effectively building a dedicated computational foundation for Codex. This indicates that OpenAI is no longer fixated on which model is smarter but is fully committed to seizing the high ground of “large-scale reliable services,” as stability and efficiency are more critical to enterprise clients than mere model intelligence.

Anthropic’s Rise: 204.9 Billion Revenue and Enterprise Coding Dominance

As OpenAI pivots, Anthropic has experienced explosive growth, doubling its annual revenue to 204.9 billion RMB, driven primarily by its enterprise coding product, Claude Code. Unlike OpenAI, Anthropic has focused on the enterprise coding sector from the start, securing numerous enterprise contracts through Claude Code’s precise adaptation. However, Anthropic faces significant challenges, including capacity limits and power issues, which hinder its competition with OpenAI.

Verdent’s Differentiation: Not Competing on Models, But as a “Multi-Model Manager”

While the two giants compete on computational power and model intelligence, Verdent has taken a completely different path—focusing on being a “model orchestrator.” By integrating Claude, GPT, Gemini, Kimi, and GLM into a single platform, Verdent allows users to freely choose from high-quality models.

Verdent’s core value lies in its orchestration layer, which includes three main functions: planning mode to break down coding tasks based on user needs, multi-model reviews for cross-evaluation of code quality, and workspace isolation for independent project management.

Common User Issues and Solutions

Many users have reported practical issues when using Verdent. Here are some common problems and solutions:

1. Sub-agent Not Working Properly (Common Error: Subagent RPC timeout after 30s)

# Fix Steps: Reload shell configuration to refresh environment variables
source ~/.bashrc # For bash users
# or source ~/.zshrc # For zsh users
# Verify if environment variables are effective
echo $PATH | grep -o '\.local/bin'
# If not effective, manually add the path
echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.bashrc
source ~/.bashrc
# Restart Verdent sub-agent service
verdent agent restart

2. Failed to Create Worktree

# Solution: Use git worktree command to create an independent work directory to avoid conflicts
# 1. Navigate to the project repository
cd your-project
# 2. Create an independent worktree (specifying branch and directory)
git worktree add ../your-project-feature feature-branch
# 3. Associate the newly created worktree in Verdent
verdent worktree add ../your-project-feature

3. Unable to Create PR After Work Completion

# Troubleshooting Step 1: Check GitHub permissions (ensure logged in)
gh auth login
# Troubleshooting Step 2: Confirm branch is committed and pushed
git add .
git commit -m "feat: Complete XX feature development"
git push origin feature-branch
# Troubleshooting Step 3: Recreate PR in Verdent
verdent pr create --title "feat: Complete XX feature development" --body "Detailed feature description" --base main --head feature-branch

Dialectical Analysis: Opportunities and Risks for Verdent

Verdent’s rise comes at an opportune time. The Financial Times reports suggest that AI model layers are gradually becoming commoditized—meaning that the intelligence gap between different models will diminish, and “how to efficiently integrate and utilize these models” will become the new competitive focus. This is Verdent’s strength, as its orchestration layer represents the most critical differentiating advantage after model commoditization.

Moreover, the competition between the two giants has created space for Verdent to thrive. An OpenAI engineer stated, “Even if our model isn’t the best, we can still provide services.” This statement emphasizes the competition in computational power rather than intelligence. For ordinary coding professionals, this means that regardless of which giant ultimately prevails, models accessed through Verdent will become cheaper and more accessible, which is undoubtedly Verdent’s greatest advantage.

However, behind the opportunities lie significant concerns. Verdent’s core reliance is on the API interfaces of major models. If OpenAI or Anthropic change their strategies to lock models into their own tools—such as making Codex the only way to use GPT coding or Claude Code the only option for experiencing Claude coding—then third-party tools like Verdent will face a fatal platform risk.

Another point to consider is that Verdent’s current core advantage is “multi-model integration,” but is this advantage irreplaceable? As the two giants continuously enhance their tool functionalities, if they launch similar multi-model integration services, will Verdent’s space for survival be significantly squeezed? This is not only a question for Verdent but a common dilemma for all third-party coding tools.

Practical Implications: What Does This Mean for Ordinary Coding Professionals?

This battle of coding tools, seemingly a contest between giants, is closely related to every coding professional. It not only addresses our core pain points but also meets our needs for flexibility and satisfaction.

Pain Point Resolution:

Previously, using coding tools meant either choosing a single model or switching between multiple tools, which was time-consuming and laborious. Verdent’s multi-model integration solves this problem, allowing one tool to handle all mainstream models, significantly improving coding efficiency. Additionally, as models become commoditized, the cost of using coding tools will continue to decrease. For instance, many professionals currently pay 683 RMB per month for CC subscriptions, and switching to tools like Verdent could save a considerable amount.

Satisfaction of Needs:

For coding professionals, we desire not only efficient coding tools but also a flexible and reliable user experience. Verdent’s orchestration layer features, such as multi-model reviews and workspace isolation, perfectly align with these needs, granting us greater autonomy in coding and enhancing code quality. Furthermore, Verdent’s multi-model approach allows us to bypass the downtime or high costs of a single platform, preventing tool-related issues from affecting our work progress.

Realization of Benefits:

Previously, the choice of coding tools often meant being tied to giants or passively accepting the limitations of a single model. The emergence of Verdent provides us with more choices, enabling us to enjoy the services of all high-quality models without being dependent on any giant. More importantly, as competition intensifies between the two giants, we can obtain higher-quality coding services at lower costs, which is undoubtedly a major benefit for all coding professionals.

However, some users have reported that Verdent still has shortcomings, such as instability in the free trial version’s sub-agent, issues when creating worktrees and PRs, and a need for further optimization in screen reader compatibility. This is a core reason many professionals hesitate between Verdent and Zenflow, as the compatibility of screen readers directly impacts the tool’s usability for users with special needs.

Interactive Topic: What Coding Tools Are You Using? What Pitfalls Have You Encountered?

The competition among coding tools is becoming increasingly fierce, with OpenAI betting on Codex, Anthropic earning 204.9 billion with Claude Code, and Verdent taking a different path with its multi-model approach. Each of the three players has its strengths and weaknesses.

Many coding professionals have experienced “tool selection anxiety”: some find the giant tools too expensive, while others feel third-party tools lack stability; some want to switch from the monthly 683 RMB CC subscription but are torn between Verdent and Zenflow; and others have encountered pitfalls with sub-agents, worktrees, and PR creation while using Verdent, leading them to abandon the tool.

Today, let’s discuss: What coding tools are you currently using? Are you sticking with the giants’ Codex or Claude Code, or have you tried third-party tools like Verdent? What difficult problems have you encountered during use? Do you think Verdent can establish itself in the gap between the two giants? Feel free to share your experiences and thoughts in the comments!