INSIDE’s brief compatibility note says Apple Intelligence support is almost equivalent to Siri AI support. However, it highlights an exception: some features need a more advanced on-device model. Those higher-end Siri AI capabilities currently support only iPhone 17 Pro, iPhone 17 Pro Max, and iPhone Air.
A LocalLLaMA user tried to benchmark Google’s new fully local dictation app, Eloquent, against open ASR models such as Qwen3-ASR and NVIDIA Parakeet V3. The tester reported that roughly half of dictations returned only fragments, even during manual use. When Eloquent produced complete transcripts, its word error rate was competitive, but the missing-output behavior made the app unreliable for evaluation and practical use.
Simon Willison says Apple’s 2024 Apple Intelligence rollout made him cautious, so he will believe the WWDC 2026 Siri AI claims only after seeing results. He notes the new features look more feasible, especially with a custom Gemini-derived model running on Private Cloud Compute. He also highlights vision LLM screen understanding and the new Core AI library for running PyTorch-derived models on Apple hardware.
Ars Technica reports that Apple is working to compress Google’s massive Gemini model so it can run on iPhone and power a new Siri experience. The short summary emphasizes a key constraint: even with on-device ambitions, a cloud component is probably inevitable. Details remain limited, so the report is best read as a signal about Apple’s AI direction rather than a confirmed product launch.
Google DeepMind published a blog post on November 20, 2025 titled "Introducing Nano Banana Pro." As the full content of the original article is not publicly…