Tech • AI • Robotics • Game
Big AI News: Opus 5.5 vs GPT-6 Sol, NotebookLM Updates, Muse Charm & More!
9.4/10TL;DR
Anthropic, OpenAI, Google, xAI, and Meta all rolled out major AI updates this week, with Claude Opus 5.5 emerging as a notably stronger all-around model than GPT-6 Soul in writing, design, and video-editing tests.
KEY POINTS
Model launches clash
Anthropic and OpenAI released new flagship models on the same day: Claude Opus 5.5 and GPT-6 Soul. In direct side-by-side use, GPT-6 Soul was generally cheaper and sometimes faster, but Opus 5.5 delivered stronger output quality on everyday tasks. The comparison centered on practical work rather than benchmark scores alone, especially writing, research, design, and editing.
Writing test favored Opus
In a browser-enabled scriptwriting test, GPT-6 Soul finished research in about 1 minute 30 seconds, while Opus 5.5 finished in a little over 2 minutes. Despite the speed edge, Opus 5.5 produced a clearer, more coherent opening and stronger narrative structure. After a week of use, it stood out as especially effective for humanlike writing and tone matching.
Design and motion graphics results split
A motion-graphics test comparing GPT-6 Soul, Opus 5.5, and Fable 5.1 put GPT-6 Soul last. Fable 5.1 produced the best result overall, while Opus 5.5 came close enough that its lower cost could make it the more practical choice for routine design work. The outcome suggests Opus 5.5 is becoming competitive outside pure text generation.
Video editing exposed a gap
In AI-assisted video editing, Opus 5.5 significantly outperformed both GPT-6 Soul and GPT-6 Astra. It handled zooms, highlights, framing, and on-screen focus more accurately, while GPT-6 outputs included awkward crops, wrong click targets, and poor highlights. Opus 5.5 completed the task in 7 minutes 15 seconds, compared with 15 minutes 30 seconds for GPT-6 Soul, while both used roughly 1% of available usage.
Opus 5.5 reshapes the model ranking
After broader testing, Opus 5.5 was judged the strongest currently available general-use model, even ahead of Fable 5.1 when balancing quality, speed, and cost. By contrast, GPT-6 Soul was viewed as underwhelming and in some cases worse than GPT-5.6. That marks a sharp turnaround for Anthropic, whose earlier Opus 5 had failed to gain traction.
Gemini Notebook adds live and interactive tools
Google expanded Gemini Notebook with live voice chat for Ultra subscribers, allowing users to question notebooks conversationally and interrupt responses. The platform also now lets notebooks be cited directly inside Google Docs through the Gemini side panel. A new interactive report format can generate embedded assets such as infographics, slide decks, mind maps, flashcards, and quizzes inside the report itself.
Google broadens Gemini ecosystem
Google also introduced a new Gemini-first Chromebook concept with built-in pointing and dictation features, positioning the device as tightly integrated with Android and system-level AI assistance. Gemini now works with 13 new apps, extending its utility across more services. The company also released Gemini 3.8 text-to-speech, an expressive audio model with realistic details such as microphone-style plosives.
xAI updates Grok and Grokbot
xAI launched Grok 4.7, describing it as a notable improvement over 4.6 at the same price and speed, though early reactions were mixed. More tangible changes arrived in Grokbot, which can now route browsing traffic through a user’s local computer to reduce blocks from websites. Voice memos and voice calls were also added, giving each bot its own selectable voice and making routine automated updates easier to consume.
Meta pushes Muse toward mainstream use
At Meta Connect, Meta made Muse the centerpiece of its AI strategy. New features include a dedicated email address for every Muse, incoming video calls with animated avatars, and Mac computer control for app-based task execution. Meta said Muse will remain largely free, with revenue focused on transaction fees when the agent completes purchases.
Shopping and hardware are central to Meta’s plan
Meta is expanding shopping integrations for Muse through Walmart, Best Buy, Sephora, Wayfair, Shop Pay, PayPal, Instacart, and Expedia. Amazon has blocked Muse access, a move that could matter if agent-based shopping becomes common. On hardware, Muse is coming to Meta AI glasses, a camera-free Ray-Ban Meta Audio model, and a compact Muse Charm device with a fingerprint sensor and camera for portable AI interaction.
CONCLUSION
This week’s releases showed the AI market moving beyond headline model launches into practical competition over writing quality, editing accuracy, voice interfaces, shopping, and hardware. The clearest takeaway was that Claude Opus 5.5 appears to have gained a real edge in day-to-day creative and productivity work.
Ask a question