™All Models & Products
-
Google Opens SynthID Detector Worldwide: Check Any Photo for AI Watermarks
Why it mattersThe Detector was previously limited to a few journalists and researchers; now anyone who signs in can upload files to check.
-
OpenAI Publishes 722 AI-Generated Math Manuscripts Citing Riemann Hypothesis Progress; None Peer-Reviewed
Why it mattersIt is the largest of OpenAI's math releases: 722 manuscripts across 372 problem families, none peer-reviewed; the internal model behind them is unreleased, and no prompts or compute costs were published.
-
Google Launches EmbeddingGemma 2: 740M-Param Multimodal Embedding Model Goes Apache 2.0, Runs On-Device
Why it mattersThe first generation passed 20 million downloads in a year; the new one puts text, code, images, audio and video in one vector space under Apache 2.0, with the text-only build at about 191MB of RAM.
-
Google Unveils Gemini 4 'Argon': Its First New-Generation Flagship Since Gemini 3
Why it mattersArgon is Google's first new-generation flagship since Gemini 3 in November 2025; Anthropic and OpenAI each also released a model priced at $2/$10 this week.
-
Nvidia-Backed Reflection Debuts Beam, Its First Open-Weight Frontier Model: 501B Params, 23B Active
Why it mattersBeam has 501B total parameters with 23B active, and the company says it needs a third to a quarter of GLM-5.2's inference compute; scores are self-reported and weights open later in October.
-
Google Trims Free Gemini: Only Flash-Lite Left From October 9
Why it mattersFrom October 9 free users get only Flash-Lite and AI Plus loses Pro, while AI Pro gains Deep Think, previously Ultra-only; Gemini reportedly has over a billion monthly users.
-
Black Forest Labs Launches FLUX 3 Image: Bounding-Box Edits Leave Pixels Outside the Box Unchanged
Why it mattersFLUX 3 Image confines edits to bounding boxes, leaving pixels outside them unchanged across multi-turn edits; it is the FLUX 3 image branch's first product release, with open weights expected in the coming weeks.
-
Microsoft Ships Its Own Voice Stack: 60-Language Streaming Transcription + Seconds-Level Voice Cloning
Why it mattersAll three transcription and speech models are built in-house by Microsoft; transcription has ~100ms first-word latency at $0.54 per audio hour, and the Flash TTS model runs at 150ms latency for $15 per million characters.
-
OpenAI Halves Its $200 Pro Plan's Usage and Drops the 5-Hour Window
Why it mattersThe price stays at $200 a month while API-equivalent usage is half the old plan's; the five-hour usage window is gone for good, and the weekly allowance can be used at any pace.
-
Anthropic Releases Claude Sonnet 5.5: Same Sticker Price, ~30% Cheaper Per Task
Why it mattersThe $2/$10 price is unchanged, but tasks use fewer tokens and output is 30%+ faster, cutting per-task cost by about a third; Box measured 2.4x speed with 12% fewer tokens.
