Google

Company-reported 87% on the embodied reasoning benchmark ERQA. Google said that from a single photo of a real tabletop Argon built a MuJoCo simulation twin on its own and collected more than 4,500 episodes, raising real-

A benchmark that separates out and measures the failures of agents working in between people as a problem of "mediated information" rather than of "ability".

Institutionalizes an access-control approach that deploys dual-use offense-and-defense models only to vetted defenders

Instead of feeding in a long video whole, the model explores it on its own — multimodal input is folded into the agent loop

Video generation moves from one-off clips to timeline work of extending and controlling

With MCP and A2A unified under a single neutral foundation, the governance of agent standards is reorganized around a foundation

The stage at which an on-device runtime covers not just mobile OSes but browsers with the same stack. Figures are the vendor's own measurements

A stable-channel case of an on-device model exposed as a web platform API rather than an app. Responsibility for model distribution and storage management passes to the browser vendor

The commercial embedding lineage officially moved past the text-only generation into a natively multimodal one

The full-scale expansion of the alternative silicon camp

On-device AI on Android evolved into a system in which a system service (AICore) distributes open models directly

The opening signal of the US camp's return to truly open releases

Recasts the optimizer as an associative memory module and proposes a sequence model that learns a self-update algorithm

Accurate infographics and multilingual text rendering redefined image generation as a reasoning task; ships with SynthID

A natively multimodal design paid off as an edge in video and document understanding

The first mega-deal for alternative silicon outside NVIDIA

Responds with a dedicated, browser-specialized model; states explicitly that it is not yet optimized for desktop OS control

Character-consistent editing reaches mass-market product quality; carries an invisible SynthID watermark

Commercialized the experimental version from March, entering the embedding API race in earnest

A milestone at which small, on-device-class models began to combine multimodality with competitive benchmark scores

Generating dialogue, sound effects and background sound together with the footage ended the silent film era

Quantization-aware training became a standard step in open-weight releases

Proposed a separate agent-to-agent layer and stated explicitly that it complements MCP

A declaration that thinking was moving from a special mode to the default

The maturing of the small, on-device route

Google's first reasoning model, taking the route of showing its thinking process to users

The second Gemma generation; Google claimed the 27B competes with models more than twice its size and the 9B outperforms Llama 3 8B

Ultra-long context left the waitlist and became an option for ordinary developers

Big Tech entered the small open-weight field, but drew criticism that the license was not Apache

Declared an approach of pretraining all modalities together from the start, rather than attaching encoders afterward

At Google I/O, Google announced over 25 products and features powered by PaLM 2, including Bard, and said the smallest size, Gecko, can run on mobile devices even offline

Emergent social behavior in a sandbox of 25 agents, produced by an observation-memory-reflection-planning architecture

Google said it would start with a lightweight version of LaMDA to reduce compute needs and gather feedback, before wider public availability in the coming weeks

Sidesteps the memory-bound nature of decoding without changing the output distribution

A technique that made embedding dimensions adjustable after the fact

The prototype of the decomposition-and-planning lineage (Plan-and-Solve, Self-Discover)

The orchestration layer that made PaLM-class training possible, and the execution foundation of the JAX/TPU stack

The starting point of the entire lineage of changing reasoning performance through prompts alone. With 8 CoT exemplars, the 540B model reached SOTA on GSM8K

The 137B model outperformed the 175B GPT-3 zero-shot on 20 of 25 tasks

Unlike most language models it was trained on dialogue, and Google said fine-tuning after training could significantly improve the sensibleness and specificity of its responses

Simplified MoE and became the standard reference for routing instability and load-balancing loss

Scaled MoE up to a 600B-scale system and established expert parallelism and automatic sharding

The common foundation shared by XLA, TVM, Triton and Mojo. Ended the practice of writing a new compiler for every accelerator

The first large-scale standard corpus built by cleaning Common Crawl with heuristics

Fine-tuned with just one additional output layer, it set new state-of-the-art results on 11 NLP tasks, including GLUE 80.4% and SQuAD v1.1 F1 93.2 (as reported by the authors)

Self-attention alone, without recurrence, made parallel training possible, establishing the architectural premise for large-scale pretraining

Disclosed the design and measured performance of the first-generation TPU, becoming the reference document for discussions of dedicated accelerators

The starting point of the ML compiler race. The layer that made TPUs practical and the foundation of all of JAX

The first demonstration, through conditional computation, of an architecture whose total parameters far exceed its active parameters; the ancestor of every MoE

The paper that turned architecture design itself into a search problem

The point at which neural machine translation crossed over from research to commercial service

A system description of attention-based seq2seq implemented at production scale

The standard technique for training on sensitive data, and a technical reference point for later regulatory discussions

The start of the path toward silicon dedicated to neural networks rather than general-purpose GPUs

Opened up the path to on-device training on mobile

Opened up an internal system, kicking off the framework competition and making deep learning tools an industry standard

Formalized distillation, which transfers a large model's knowledge to a small model. The root of today's small-model lineage

Stabilized and sped up the training of deep networks and was adopted as standard in most architectures

Formalized the encoder-decoder that maps a sequence to a sequence. The ancestor of today's generative model interface

Laid the organizational foundation that would later produce DQN and AlphaGo

The starting point of every dense retrieval and embedding stack that followed

Separate industry groups confirmed the same shift (HMM→DNN), and deep learning became the industry standard for speech recognition

Showed that high-level concepts emerge from nothing more than large-scale unlabeled data and large-scale compute. A symbol of Google's investment in deep learning