SWE-Bench | byteiota

Tag: SWE-Bench

Google EnvHarness open-source adaptive AI agent training environment diagram

News

Google EnvHarness: Open-Source Agent Training That Adapts to Your AI

Google open-sourced EnvHarness on Oct 2: a wrapper that reshapes agent training benchmarks without touching ...

By ByteBot

3 days ago

AI & Development

ProgramDistill: AI Coding Agents Still Fail Half the Time

ProgramDistill exposes a critical gap: even GPT-6 Astra fails 51% of real-world app reconstruction tasks. ...

By ByteBot

September 28, 2026

Claude Opus 5.5 neural network nodes with version number highlighted in blue

News

Claude Opus 5.5 Is the New Default — Here Is What Changes

Anthropic’s Claude Opus 5.5 is now the default in Claude Code. It’s 40% cheaper per ...

By ByteBot

September 28, 2026

OpenHands 1.0 autonomous coding agent running inside a secure Docker container with blue security shields

AI & Development

OpenHands 1.0: Self-Hosted Coding Agent With Safety Sandbox

OpenHands 1.0 ships production-grade Docker sandboxing and hits 68% on SWE-bench Verified. Here is what ...

By ByteBot

September 9, 2026

Bold conceptual visualization of Tencent Hy4 770B model with benchmark scores and data particles on blue and white background

News

Tencent Hy4 Open-Weights Model: Benchmarks and Access

Tencent released Hy4 preview August 28: 770B open-weight model, Apache 2.0, free for 2 weeks. ...

By ByteBot

August 30, 2026

Microsoft Agent Lightning v1.0 featured image showing AI agent training with reinforcement learning concept

AI & Development

Microsoft Agent Lightning v1.0: Train AI Agents Without Code Rewrites

Microsoft Agent Lightning v1.0 trains your existing LangChain or AutoGen agent via reinforcement learning with ...

By ByteBot

August 30, 2026

Data visualization dashboard showing Ornith-1.5 Terminal-Bench benchmark scores versus Claude Opus 4.8, with self-improvement loop diagram

News

Ornith-1.5 Closes the Self-Improvement Loop on Open Source

DeepReinforce shipped Ornith-1.5 this week, and the Hacker News crowd landed on it fast. The ...

By ByteBot

August 19, 2026

Kimi K3 model available in GitHub Copilot model picker interface

AI & Development

Kimi K3 Is Now in GitHub Copilot: What the Open 3T Model Means

Moonshot AI’s Kimi K3 — a 2.8T open-weight model — is now in GitHub Copilot’s ...

By ByteBot

August 11, 2026

Microsoft Orchard open-source agentic AI training framework diagram showing Kubernetes-native environment service with SWE, GUI, and Claw agent domains

AI & Development

Microsoft Orchard: 73% SWE-bench With 3B Parameters

Microsoft Research open-sourced Orchard, a Kubernetes-native agent training framework that hits 73% on SWE-bench Verified ...

By ByteBot

August 7, 2026

GLM-5.2 neural network visualization with benchmark data and code snippets on dark blue background

AI & Development

GLM-5.2 Beats GPT-5.5 on SWE-bench — And You Can Self-Host It

Z.ai’s GLM-5.2 outscores GPT-5.5 on SWE-bench Pro with MIT-licensed open weights at 6x lower cost. ...

By ByteBot

August 3, 2026

12 3...5

Tag: SWE-Bench

Posts navigation

Categories

Latest Posts