Bullet is hitting 95.8% on SWE-bench Verified and it's way faster

Bullet, an AI coding agent developed by Code with Bullet, has achieved a 95.8% score on SWE-bench Verified, outperforming existing tools like Claude Code while reducing latency through smart model routing, targeted search, context hygiene, and parallel investigation, which cut round trips by 16% and costs by 27%.

Bullet is hitting 95.8% on SWE-bench Verified and it's way faster The founders basically built this because they were sick of the latency and inefficiency of existing tools like Claude Code /en/tags/claude%20code/ . Instead of just hoping for a faster model, they focused on the AI workflow and how the agent actually interacts with the codebase. How they actually optimized the agent loop If you're into prompt engineering or building LLM agents, the technical choices they made to kill latency are worth noting: 1. Smart Model Routing: They stopped blindly sending every task to the heaviest model. If Sonnet can handle it, it goes to Sonnet, preventing unnecessary compute lag on simpler tasks. 2. Targeted Search over Embeddings: Rather than embedding the entire repository—which can be noisy—or stuffing a massive compressed context into the window, they implemented high-speed, targeted greps to find exactly what's needed. 3. Context Hygiene: This is a huge one for avoiding "model drift." They bound tool outputs and purge stale screenshots or redundant file reads so the context window doesn't get flooded with garbage. 4. Parallel Investigation: Instead of a linear "search - read - edit - verify" loop, Bullet batches independent investigations. It gathers all necessary info in parallel, makes one surgical edit, and then does a single focused verification. This reportedly cut round trips by 16% and costs by 27%. Real-world friction and findings The development process revealed some annoying "silent killers" in AI coding. For instance, regex-dialect mismatches in code search can lead an agent down a completely wrong path without the model even realizing it missed the target. To fix this, they had to build a search system with specific fallbacks and bounded context to ensure the agent stays on track. The most practical use case for this kind of speed isn't just small bug fixes, but long iterative workflows—things like data pipelines or evaluation loops where each step relies on the previous one. In those scenarios, you can't just spin up ten parallel agents; you need a single agent that can move through the sequence as fast as humanly possible. If you want to see the specific benchmarks or the methodology they used to hit those numbers, they've detailed it here: https://www.codewithbullet.com/blog/benchmark-results.html For a visual on how the agent handles a live codebase, there's a demo available: https://www.youtube.com/watch?v=rWVmG5fRKgE Bullet just hit 95. 7h ago /en/news/6158/ AI is making code reviews the new bottleneck in the SDLC 19h ago /en/news/6084/ Immigration lawyers can't just blindly trust AI for case filings 1d ago /en/news/6053/ The AI era is wiping out the mid-level software engineer 1d ago /en/news/6032/ Anthropic Claude Code might have a security backdoor according 1d ago /en/news/5946/ AI Pulse adds a fake LED strip to the macOS Dock for agent status 2d ago /en/news/5863/ Next Microsoft is finally killing the double Copilot icon clutter → /en/news/6195/ these real-world AI monetization case studies https://tanyan888.com/ , with plenty of directly applicable cases. All Replies （4） aria-label="Hidden secret code" ? Lol. It's a fun little mini game, but the accessibility tags totally gave it away.