48 stories tagged by Digg
AI
A user says they asked Opus 5.5 to choose a text from their library to turn into a movie. They describe the chosen work, ‘The Long Light,’ as a world by Claude Fable 5.
AI
A user calls it “wild” that Opus 5.5 appears to know how to use Runway, saying it takes just one prompt.
AI
A user reports that the two models ran fastest with different setups in a multi-agent evaluation on ProgramBench.
AI
A post says tests of the latest AI models on 2,400 engineers led its team to name Opus 5.5 its new daily default. It describes GPT-6 Luna as 20 times cheaper than Opus 5.5 for high-volume routing.
AI
An early tester says Sonnet 5.5 is much faster and 50% cheaper than Opus 5.5. He calls his experience with it “incredible.”
AI
A user calls a choice by Opus 5.5 unprompted and “bold,” but doesn’t explain what the choice was in the post’s text.
AI
A user wrote that Claude “suddenly stopped cheating” but did not explain in the post text what “cheating” meant.
AI
A user says a shared result was made pretty much in one shot with Opus 5.5. They also say AI models are getting very good at 3D visualization and technical explanations.
AI
A user says they generated an animation video for @animosapp without touching a single keyframe.
AI
A user describes a video comparing GPT-6 Astra, Opus 5.5, Grok 4.7 and MolmoAct2 as unsped and not cherry-picked. They say LLM robot control is progressing but has a way to go.
AI
The user who ran the test says Opus 5.5’s physics and visual quality wasn’t noticeably worse. They suggest line count could be a rough measure of code elegance, though not a perfect one.
AI
Bend 2’s creator says they asked Opus 5.5 to make an intro for the programming language, then shared a video with the reaction: “WTF is this?”
AI
A user says Opus 5.5 would not transcribe a photo of a page from a 1960 book they own, attributing the refusal to copyright.
AI
A user says they let the model choose the subject, then it wrote the lyrics and drew every frame in code using real NASA data.
AI
CancerBench's creator says the two models have been added and now tie for first and last place. The benchmark's only metric counts how many types of cancer a model has cured.
AI
A user says Opus 5.5 on a $100 Claude plan lasts them six to seven days a week, compared with one to 1.5 days for Astra on a $200 Codex plan.
AI
A user says Opus 5.5 has a penchant for suddenly cutting the music and video to black for drama.
AI
A post sharing ‘Loom Jockey’ links to a video and gives separate credits for its lyrics, music and video.
AI
A user imagines GPT-6 Terra launching at $1/$5 and outperforming Opus 5.5 on everything.
AI
A post says the combination picks a model based on the prompt and can set up complex multiplayer apps.
AI
A user shared the video and described the poems as ‘divined by 405base.’
AI
A user hopes “it” isn’t Opus 5.5 again, but suspects it probably will be for at least some queries.
AI
A user says Claude Opus 5.5 handled difficult programming tasks and video creation impressively, calling it a leap over Astra.
AI
A user says they had Opus 5.5 build a player and tools for Another World’s approach to animation. They say the video’s bin and data came in under 250KB with crude compression; the audio is larger.
AI
A post says Claude Opus 5.5 uses em dashes about 95% less often. Citing Arena analysis, it also says the model writes sentences that are 17% shorter.
Technology
A user calls this combination of Claude tools “pretty amazing.”
AI
A user says something they quoted “should be a model eval,” without describing what it would test.
AI
A user says the build has 277K logic gates, an operating system and games. They say it ran autonomously from one prompt over about 30 hours in ultracode mode.
AI
A user describes Opus 5.5’s subagent orchestration in Amp, highlighting its polite-sounding messages.
AI
A user worries that people could offset AI agents' productivity gains by becoming lazier. A reply calls for a “Claude Code for self development.”
