Latest

The split matters more than the tie: Anthropic models dominate completeness, with Claude Sonnet 5.5 at 64.8% versus 42.2% for the best non-Anthropic model, while OpenAI models dominate accuracy, led by GPT-6.1 Sol at 96.7%. Teams deploying models for medical record review should pick based on which failure mode they can tolerate. Even top models still fail more than half of the benchmark's questions, so human review remains necessary.

官方账号Artificial Analysis@ArtificialAnlys11 sourcesSource

Legacy COBOL systems still run banking and government payroll worldwide, and a 97.9 percent migration accuracy claim suggests AI-assisted mainframe modernization is becoming practical rather than experimental. Teams maintaining COBOL or Fortran codebases should evaluate GPT-6.1 Sol against Anthropic's Claude and other coding assistants on their own migration test suites. The MultiLangBench scores also give enterprises a benchmark for comparing models beyond mainstream languages like Python.

官方账号Decoder@Matthias Bastian11 sourcesSource

The price of intelligence is collapsing at every fixed quality level while top-end capability prices keep climbing, so buyers face a widening spread between commodity and frontier performance. Teams benchmarking models should match their task's required score to the cheapest threshold model instead of defaulting to the highest scorer. The data also names concrete price leaders at each level, such as Claude Haiku 5.5 (xhigh) at US$0.12/task for a score of 41.

官方账号Artificial Analysis@ArtificialAnlys11 sourcesSource

The index measures failure modes most leaderboards ignore: unauthorized actions, false attribution and deceptive completion, so lab rankings can shift even when capability scores stay flat. Reflection AI debuting as the top US open model on Agent Arena gives open-weight builders a concrete alternative to closed systems. StepFun's entry at 75.1%, nearly matching Meta on unauthorized actions, is a datapoint worth watching for anyone tracking Chinese labs against US peers.

lmarena.ai@lmarena_ai11 sourcesSource

Benchmarks that pause the world reward slow, careful reasoning and hide latency penalties, so leaderboard rankings may invert under realtime conditions. Developers building latency-sensitive agents can copy the TORSI pattern: keep a fast model in the live loop and let a stronger model write and test its instructions offline between rounds. The result applies to any realtime control task, not just games.

rohanpaul_ai@rohanpaul_aiSource

The benchmark's merge-focused grading penalises the messy diffs and unhelpful tests that cheaper evaluations ignore, so leaderboard gaps here reflect real maintainer acceptance rather than raw test-passing. Leading models still fail over 40% of tasks, and nine tasks went unsolved by every model evaluated. Teams choosing coding agents can weigh GPT-6.1 Sol's 49% score at US$0.81 per task against Claude Opus 5.5's 52% at far higher cost, with open-weights options like MiMo-V2.6-Pro at 45% close behind.

官方账号Artificial Analysis@ArtificialAnlys11 sourcesSource

The workflow itself is the lesson: pairing a reverse-engineering tool like REA with a strong coding model can reproduce closed product features as open, inspectable skills, without ever copying the original prompts. Developers building agent tools can follow his five-step process with Codex and Claude Code 5.5, or just try the design skill in BaoCut locally. BaoCut is open source on GitHub, so the skill and frontend code are fully viewable.

宝玉@dotey9 sourcesSource