SWE-bench Verified
PulseAugur coverage of SWE-bench Verified — every cluster mentioning SWE-bench Verified across labs, papers, and developer communities, ranked by signal.
- instance of Terminal-Bench 2.1 90%
- instance of Terminal-Bench V2 90%
- instance of SWE Bench Pro 70%
- used by Terminal Bench 2.0 70%
- instance of Apache Software License 2.0 70%
- instance of SWE-Bench Multilingual 70%
- used by DagsHub 70%
- used by vLLM 70%
- competes with Terminal Bench 2.0 70%
- affiliated with SWE Bench Pro 70%
- instance of GLM-5.2 70%
- used by Ollama 60%
13 day(s) with sentiment data
-
Visual code representations show mixed results for AI coding agents
A new study explores the use of rendered code as visual representations for AI coding agents, aiming to reduce token costs and improve repository-level issue resolution. The research found that while visual code can dec…
-
New tool PAIChecker identifies PR-issue misalignment in LLM benchmarks
Researchers have developed PAIChecker, a multi-agent system designed to identify and correct misalignments between pull requests (PRs) and their associated issues in benchmarks used to evaluate large language models (LL…
-
DeepSeek V4-Pro leads open-source coding LLMs with strong benchmark performance
DeepSeek V4-Pro has emerged as a top-tier open-source coding LLM, achieving an impressive 80.6% on the SWE-bench Verified benchmark. While this model requires significant server infrastructure, other powerful open-sourc…
-
Tsinghua University releases VeriLoop Coder-E1 for verifiable code repair
Researchers from Tsinghua University have open-sourced VeriLoop Coder-E1, a model designed for verifiable recursive self-improvement in code repair. Built upon the Qwen3.6-27B architecture, VeriLoop Coder-E1 utilizes an…
-
Vendors clash over 'Agentic Development Lifecycle' definitions
The concept of an "Agentic Development Lifecycle" (ADLC) is emerging as a response to the unique challenges posed by AI systems, which break traditional software development assumptions. However, there is significant di…
-
Thinking Machines releases Inkling-Small, outperforming larger predecessor
Thinking Machines Lab has launched Inkling-Small, a new open-weights multimodal model that prioritizes efficiency over sheer size. Despite being significantly smaller than its predecessor, Inkling, Inkling-Small demonst…
-
DarwinX system uses natural selection to evolve LLM agent harnesses
Researchers have introduced DarwinX, a novel system that employs natural selection principles to evolve the "harnesses" of large language model (LLM) agents. Instead of modifying model weights, DarwinX focuses on optimi…
-
NVIDIA unveils NOOA framework for AI agents using Python objects
NVIDIA has introduced NOOA, a new framework for building AI agents that utilizes Python objects as a core abstraction. This approach aims to consolidate agent development, which is typically spread across prompt templat…
-
Claude Code vs. Cursor: AI coding tools compared for developers · 6 sources tracked
Developers are comparing Claude Code and Cursor, two AI coding tools, with distinct strengths and weaknesses. Claude Code, a terminal-native agent, excels at complex, multi-file refactors and autonomous planning due to …
-
LLM attention patterns linked to automated program repair success
A new empirical study investigates how Large Language Models (LLMs) process bug reports for automated program repair. Researchers analyzed attention patterns in LLMs across 319 Python and Java bugs from SWE-bench Verifi…
-
Alaya Token integrates Kimi K3, world's first 3T open-source model
Alaya Token, a platform from DataCanvas, has integrated Kimi K3, the world's first open-source 3 trillion parameter model. This integration allows users to access Kimi K3's advanced capabilities, including its 1 million…
-
MindForge pipeline trains small LLMs for full software engineering lifecycle
Researchers have developed MindForge, an automated pipeline designed to train smaller language models in comprehensive software engineering tasks. This system converts open-source command-line programs into source-free …
-
Anthropic's Claude Opus 5 tops leaderboards, but users debate value and guardrails
Anthropic has released Claude Opus 5, which has achieved top rankings on several AI leaderboards, including SWE-bench and FrontierBench. A key innovation is the introduction of an 'effort' parameter in the API, allowing…
-
New CoHarden framework improves AI-driven bug fixing with iterative test hardening
Researchers have developed a new framework called CoHarden to improve the effectiveness of automated program repair (APR) using large language models (LLMs). The framework addresses limitations in existing bug reproduct…
-
Anthropic's Claude 3.5 Sonnet enhances coding, while GPT-4o faces multimodal input issues
Anthropic has released Claude 3.5 Sonnet, a new coding-focused LLM that boasts a 49.0% improvement on the SWE-bench Verified benchmark. This model is designed to reduce hallucinations and enhance developer workflows thr…
-
New SWE-Pruner Pro method optimizes coding agent context by 39%
Researchers have developed SWE-Pruner Pro, a novel method for efficiently managing long contexts in coding agents. Unlike previous approaches that used separate classifiers, SWE-Pruner Pro leverages the agent's internal…
-
Anthropic, OpenAI, and Next.js launch major updates in April 2026
The web development landscape saw significant shifts in April 2026 with major AI model releases and tool updates. Anthropic launched Claude Opus 4.7, which now leads on the SWE-bench Verified benchmark, while OpenAI rel…
-
Coding agents need minimal context for editing, study finds
A new research paper investigates the minimal context required for coding agents to effectively edit code. The study found that natural language summaries of code are largely ineffective for resolving issues, performing…
-
AI coding agent token reduction may not cut costs, research finds
A new paper argues that reducing tokens in AI coding agents does not necessarily reduce costs, and can even harm task completion. The research found that prompt-cache traffic significantly contributes to overall costs, …
-
UMoE pipeline enhances domain-specific MoE model training
Researchers have introduced UMoE, a novel pipeline designed to optimize Mixture-of-Experts (MoE) models for domain-specific tasks. This method involves pruning underperforming experts, regrowing the expert pool to its o…