#LM Arena
257 related articles
·5 min
Gemini 3.5 Pro Delayed Again: Google's Compute Dilemma and the Mystery of a Full Rebuild
Gemini 3.5 Pro is delayed again, yet the community's reaction is surprisingly calm. This article dives deep into Google's compute cost pressures, the risks of a full architectural rebuild, and DeepMind's long-term strategy.
Read more →
·3 min
DeepSeek-V4 Flash RL Training Successfully Ported to AMD MI355X GPU: A Complete Breakdown
The Miles team and AMD announce the full port of DeepSeek-V4 Flash RL training to AMD Instinct MI355X GPUs on ROCm, boosting AIME pass@1 from 0.39 to 0.49—a milestone for compute ecosystem diversity.
Read more →
·3 min
GPT-5.6 Released: Sol/Terra/Luna Three-Tier Model Family Takes On Claude Head-On
OpenAI officially releases GPT-5.6 with a three-tier model family—Sol, Terra, and Luna. Flagship Sol beats Claude on coding benchmarks: twice as fast, a third cheaper.
Read more →
·4 min
No Competitive Programming Required: How CS Students Can Pivot to AI/ML and Land Internships
How can CS students who dislike competitive programming systematically pivot to AI/ML? This guide covers skill priorities (Python/SQL/ML/deployment), portfolio strategy, Kaggle tips, and real paths to landing AI/ML internships.
Read more →
·2 min
In-Depth Analysis of GPT Live: How Full-Duplex Voice AI Truly Achieves Human-Like Conversation
GPT Live uses a full-duplex architecture with GPT 5.5 reasoning, enabling simultaneous listening and speaking, proactive engagement, contextual memory, and real-time translation. Voice AI moves from mechanical responses to human-like interaction.
Read more →
·6 min
1X NEO Robotic Hand and GPT-5.6: A Full Breakdown of Dual Breakthroughs in Robot Hardware and Software
1X releases a new robotic hand for the NEO humanoid robot—25 DOF, force transparency, and tactile skin enabling data self-labeling. OpenAI launches the three-tier GPT-5.6, boosting coding and cost-efficiency. Hardware and AI brains evolve together, accelerating humanoid robot commercialization.
Read more →
·4 min
Four Major LangChain Updates: OpenWiki, Voice Agents, and Sub-Agent Orchestration Explained
LangChain releases four major updates: OpenWiki for auto-generating codebase docs, voice agent tutorials, Harbor evaluation integration, and deepagents programmable sub-agents.
Read more →
·7 min
Building AI Agents in Rust: The GAIA Benchmark Reveals the Capability Ceiling of Large Models
A Rust-based AI Agent evaluation framework uses the GAIA benchmark to compare GPT, Claude, DeepSeek and other models with no tools. Results show pure LLMs cap at ~25% accuracy, revealing why tool use is decisive for Agents.
Read more →
·5 min
AI Content Floods Social Media: Why Has LinkedIn Become Ground Zero?
Generative AI tools have flooded social media with AI-generated content, hitting LinkedIn hardest due to its professional nature. A deep dive into the causes, ecosystem impact, and solutions.
Read more →
·7 min
Grok 4.5 Launch: Coding AI Market Reshuffle, Anthropic Claims Half the Territory
Grok 4.5 launches at just $0.49 per task, 90% cheaper than rivals. Anthropic's Claude Code claims 50% of the AI coding market. SambaNova raises $1B. Read the latest AI market shifts.
Read more →
·3 min
GPT-5.6 Released: Codex Takes Over ChatGPT, Ushering in the Era of AI Agents
OpenAI releases the GPT-5.6 series, deeply merging ChatGPT with Codex as the Agent engine takes over the underlying architecture. A deep dive into its impact on AI coding toolchains and Agent applications.
Read more →
·5 min
Getting Started with AI Agents: Concepts, Architecture, and Core Components Explained
What is an AI Agent? Starting from Bill Gates' claim about the computing revolution, this article explores AI Agents' intuitive concepts, four core components (LLM+Planning+Memory+Tools), and what Agent development means for programmers.
Read more →
·2 min
OpenAI's Four-Front Offensive: A Deep Dive into GPT-5.6, Codex, Voice AI, and Work Agent
OpenAI launches GPT-5.6, ChatGPT Work, upgraded Codex super-app, and GPT Live voice AI — a deep dive into all four products and their impact on the AI landscape.
Read more →
·3 min
Java Developers Pivoting to AI: Why the Agent Direction Is the Best Choice
How can experienced Java and backend developers pivot to AI? This deep-dive explains why the Agent direction is the best fit — skills transfer well, market demand is high, and the path from "using frameworks" to "understanding source code" is clear.
Read more →
·5 min
GPT-5.6 Trio Released: A Deep Dive into Sol/Terra/Luna and the Breakthrough in Design Judgment
OpenAI officially releases the GPT-5.6 series with three models: flagship Sol, balanced Terra, and economy Luna. A deep dive into its core breakthroughs—a step change in design judgment and enhanced computer-use capabilities.
Read more →
·5 min
Real-World Comparison of 6 Open-Source LLMs for On-Premise Deployment: A Selection Guide to Avoid Pitfalls
A hands-on comparison of 6 open-source LLMs (DeepSeek, Qwen3, Zhipu GLM, Kimi K2, MiniMax M3, Tencent Hunyuan 3) for on-premise deployment—covering hardware cost, inference efficiency, and deployment difficulty.
Read more →
·2 min
OpenAI Abandons SWE-Bench Pro: The Trust Crisis in AI Coding Benchmarks and What Comes Next
OpenAI has dropped SWE-Bench Pro as a recommended AI coding benchmark, exposing deep issues like data contamination and metric limitations. We explore the trust crisis and where evaluation is headed.
Read more →
·6 min
Tencent Hunyuan HY3 Official Version Open-Sourced: The Strongest Under 300B, at 1 Yuan per Million Tokens
Tencent Hunyuan HY3 official version is open-sourced under Apache 2.0, priced as low as 1 yuan per million input tokens, with major gains in agents, reasoning, coding, and long context. On the same day, Meituan open-sourced its trillion-parameter LongCat 2.0.
Read more →
·3 min
Huawei Pangu 2.0-Flash Benchmark Review: #1 in Instruction Following, Last in Engineering Code
Huawei OpenPangu 2.0 Flash review: 92B MoE open-source model tops instruction following at 95.9, excels in math & Agent tasks, but scores last on SWE-Bench engineering code at 63.1.
Read more →
·2 min
GLM-5.2 Spotted in Testing | Claude Fable 5 Launches: 4 Key AI Updates
GLM-5.2 spotted in testing, Anthropic launches Claude Fable 5, Moore Threads open-sources MusaCoder for domestic GPUs, and Google releases Gemini real-time translation.
Read more →