Nano Banana 2 spreads faster image editing across Google’s apps
Google keeps Nano Banana Pro available for eligible paid users.
Capvise Media / Stories
AI
The systems learning, generating and acting around us.
90 storiesBrowse by month
That coding AI score has more behind it than the model
Infrastructure limits, flawed tests and prior exposure can all change what a score means.
Gemini 3.1 Pro reaches developers in preview
Google rolled out access across AI Studio, Vertex AI and consumer services.
Gemini adds 30-second music generation with Lyria 3
The beta accepts text or images and embeds SynthID in generated audio.
Sonnet 4.6 becomes Claude’s default on Free and Pro plans
Anthropic changed the default model while keeping Sonnet’s published token rates unchanged.
Google updates Deep Think and opens an API early-access route
Ultra subscribers received the reasoning mode while selected researchers could request API access.
GPT-5.3-Codex expands the coding agent’s remit
Paid Codex access came first; API access was still being prepared.
Opus 4.6 adds a larger context option and effort controls
Anthropic kept base token prices while charging more for longer prompts.
Voxtral Transcribe 2 splits batch transcription from live speech
Thirteen languages include Hindi, with timing controls for streaming use.
OpenAI’s Codex app puts parallel coding tasks in a desktop interface
Built-in worktrees separate agents working in the same repository.
Project Genie opens interactive world generation to US Ultra users
Google limited the initial experience to adults on its highest subscription tier.
Mistral Vibe 2.0 adds custom agents and configurable modes
The release also changed Devstral 2’s standard API access to paid usage.