Claude Opus 5 Builds Full Games

PLUS: A German court rules Suno broke copyright law, and DeepSeek ships a bargain-priced agent model

Author

Gen AI Art
August 03, 2026 • Estimated Reading Time: 3 minutes

Sponsored by

Claude Opus 5 can now build a fully playable 3D game from a single prompt — geometry, physics, textures, and music, all generated as code with zero external assets. One developer’s browser-based shooter ran 55,000 lines of self-generated Three.js; another’s kart racer used 127 AI agents across 11 rounds to reach the finish line.

Games used to require armies of artists and engineers. Now the question is how far this goes: does “prompt-to-game” become a real production shortcut, or does the shine wear off once players expect more than a tech demo?

Today in AI:

  • Claude Opus 5 builds playable 3D games from prompts

  • Suno loses landmark copyright case in Germany

  • DeepSeek’s V4-Flash brings frontier agents for pennies

The best voice models now listen, adapt, and resolve too.

Most CX platforms don't own the voice. They orchestrate a workflow, then call a third party for speech and transcription. Every hop adds latency, and latency is what turns a frustrated customer into a churned one.

ElevenAgents is the opposite. Built on the voice models the market already builds on, it runs voice, transcription, chat, and reasoning in one vertically integrated pipeline. Responses come back in under 400 milliseconds and sound human, not synthetic. When a caller gets frustrated, the agent detects it and shifts tone in real time: calm, reassuring, patient.

You keep full control. Plug in any LLM, connect tools, webhooks, and MCP servers, and ground every answer in your knowledge base. Launch in minutes, A/B test with Experiments, enforce Guardrails, and version every change.

More resolved conversations, less infrastructure stitching. Pricing is transparent and flat at $0.08 per minute.

What’s new? Anthropic’s Claude Opus 5 is generating fully playable 3D games from a single text prompt, writing every line of geometry, physics, and music as procedural code that runs directly in the browser, The Decoder reports.

What matters?

  • Matt Shumer’s Three.js shooter “Claude-of-Duty” runs on 55,000 lines of self-generated code, with textures, meshes, and sound built entirely at runtime.

  • Developer Ryan Campbell’s kart racer used 127 AI agents across 11 rounds of iteration to produce a playable demo with weather, lighting, and full controls.

  • Builder Pankaj Kumar generated a Minecraft-style clone with an infinite procedural world and 15 biomes, spending 25 million tokens on the build.

Why it matters?

Unlike video-generation models that output a flat clip, Opus 5 writes editable code — meaning every texture, sound, and physics rule a game needs comes from the prompt itself, no asset pipeline required. That turns “build me a game” from a novelty demo into something close to an actual production shortcut, and the prompts behind these builds are already circulating for anyone to copy.

GUIDE

Suno Loses Landmark Copyright Case in Germany

What’s new? A Munich court ruled that Suno violated copyright law by training its AI on songs represented by German rights group GEMA, Music Business Worldwide reports — the first major loss in a wave of lawsuits facing AI music generators.

What matters?

  • GEMA, which represents roughly 95,000 members, proved its case by showing Suno’s outputs closely matched songs including “Rasputin,” “Daddy Cool,” and “Forever Young” in melody and rhythm.

  • The Munich Regional Court ordered Suno to disclose revenue tied to the infringing songs and pay damages, with the exact amount still to be determined.

  • Suno disagrees with the ruling and is weighing an appeal, while it continues fighting separate suits from Universal Music Group, Sony Music, and Denmark’s Koda.

Why it matters?

The ruling forces AI music companies to actually license the catalogs they train on instead of asking forgiveness later, and Suno’s $5.4 billion valuation just got a lot more expensive to defend. Every other AI audio generator watching this case now knows a courtroom, not a settlement, can decide how this plays out.

SPONSORED BY WISPR FLOW

10x the context. Half the time.

Speak your prompts into ChatGPT or Claude and get detailed, paste-ready input that actually gives you useful output. Wispr Flow captures what you'd cut when typing. Free on Mac, Windows, and iPhone.

What’s new? DeepSeek quietly updated V4-Flash, re-training its agent-focused model to push benchmark scores well past its previous preview version while keeping the same architecture and size.

What matters?

  • Input tokens cost $0.14 per million on a cache miss (just $0.0028 per million on a cache hit), with output priced at $0.28 per million — a fraction of what mainstream frontier models charge for comparable agent work.

  • The model scores 82.7 on Terminal-Bench 2.1 and 54.4 on DeepSWE, benchmarks that measure how well an AI agent handles real coding and terminal tasks.

  • Upgrading takes no new deployment work — developers just swap the model parameter to deepseek-v4-flash and keep their existing API integration.

Why it matters?

Cheap, capable agents change the math on what’s worth automating, and every premium provider now has to justify its price against a model that scores this close for pennies on the dollar. The next wave of AI tools built on agents will increasingly run on whichever model is cheap enough to use at scale.

Everything else in AI

Moonshot AI released weights for Kimi K3, a 2.8-trillion-parameter open model with a 1-million-token context window that trades blows with Fable 5 on benchmarks.

FCC banned imports of new foreign-made humanoid robots and robot dogs, a move aimed squarely at Chinese manufacturers.

Midjourney acquired astrology app Co-Star, with founder Banu Guler joining the company as chief design officer.

ExplainX breaks down the 10 best Claude Opus 5 game-building prompts, with copyable templates and the demos each one produced.

God of Prompt bundles its full library of prompts, guides, and AI tools into one pack for readers who want a shortcut to better outputs.

Essential AI Guides - Reading List:

Let us know!

Work with us

Reach 100k+ engaged Tech Professionals, Engineers, Managers and decision makers. Join brands like MorningBrew, HubSpot, Prezi, Nike, Ahref, Roku, 1440, Superhuman, and others in showcasing your product to our audience. Get in touch now →