Skip to main content
ForgeCode ranks #1 on TermBench with 81.8% accuracy.Learn more →
All
Benchmarks Don't Matter — Until They Do (Part 2)ForgeCode now reaches 81.8% on TermBench 2.0 with both GPT 5.4 and Opus 4.6. The interesting part is not the score. It is what we had to change in the agent to make GPT 5.4 behave as reliably as Opus 4.6.
Tushar
Tushar
Benchmarks Don't Matter — Until They Do (Part 1)ForgeCode hit 78.4% SOTA on TermBench 2.0 with gemini-3.1-pro-preview. This is the technical account of how we got there: seven failure modes, their fixes, and why the benchmark work generalized across models rather than overfitting to one run.
Tushar
Tushar
ForgeCode v0.106.0 Release: Plan Progress Tracking and Reliability ImprovementsForgeCode v0.106.0 introduces plan progress tracking for better task management and reliability improvements to enhance your development workflow.
ForgeCode Team
ForgeCode Team
Coding Agents Showdown: VSCode Forks vs. IDE Extensions vs. CLI AgentsThe AI coding assistant landscape is fragmenting into three distinct ways to integrate AI into your development workflow. Here's an objective analysis of what each approach reveals about the future of software development.
Tushar
Tushar
Claude Sonnet 4 vs Kimi K2 vs Gemini 2.5 Pro: Which AI actually ships production code?I ran Claude Sonnet 4, Kimi K2, and Gemini 2.5 Pro on the same Next.js app and measured cost, speed, and whether the code actually shipped without follow-ups.
Amitesh Anand
Amitesh Anand
Graduating from Early Access: New Pricing Tiers Now AvailableHow our explosive early access growth shaped our pricing strategy and what's now available for developers at every scale.
Tushar
Tushar
Kimi K2 vs Grok 4: Which AI Model Codes Better?A deep dive into Kimi K2 and Grok 4 for real-world coding, comparing their performance across bug fixing, feature implementation, tool use, and cost efficiency. See which model stands out and when to choose each for your dev workflow.
Shrijal Acharya
Shrijal Acharya
Kimi K2 vs Qwen-3 Coder: Testing Two AI Models on Coding TasksI tested Kimi K2 and Qwen-3 Coder on 13 Rust development tasks across a 38k-line codebase and 2 Frontend refactor tasks. The results reveal differences in code quality, instruction following, and development capabilities.
Tushar
Tushar
ForgeCode Performance RCA: Root Cause Analysis of Quality Degradation on July 12, 2025A detailed root cause analysis of the ForgeCode AI coding assistant's quality degradation incident on July 12, 2025, including the impact of aggressive conversation compaction and steps taken for future prevention and stability improvements.
Tushar
Tushar
Grok 4 Initial Impressions: Is xAI's New LLM the Most Intelligent AI Model Yet?A deep dive into Grok 4's benchmarks, architecture, and community impressions. Is xAI's latest LLM a breakthrough towards AGI, and is it worth integrating into your AI development workflow?
Arindam Majumder
Arindam Majumder
Claude 4 Opus vs Grok 4: Which Model Dominates Complex Coding Tasks?I pitted Claude 4 Opus against Grok 4 in a series of challenging coding tasks. The results highlight trade-offs in speed, cost, accuracy, and frustration factors that every dev should know.
Tushar
Tushar
ForgeCode v0.98.0: Integrated Authentication and Developer Experience ImprovementsForgeCode v0.98.0 release brings browser-based authentication, AI safety limits, and enhanced file operations for AI coding assistants. Streamline your terminal development workflow with improved reliability and developer experience.
ForgeCode Team
ForgeCode Team