Google received the first reply from the Project Suncatcher prototype satellite, with the on-orbit data-center-class TPU successfully running the open Gemma 4 model and returning a haiku about space. The satellite was launched on October 1, and the host and TPU tray have previously completed power-on and early machine learning tests. Next, Google will test how the TPU performs under orbital thermal stress and radiation exposure, paving the way for the 2027 two-satellite interconnection goal.
Google's Notebook now offers hundreds of textbooks for free; students can browse, select, and upload them directly as knowledge base sources for exam-prep questions. Partner publishers include S Chand Publishing, OpenStax, and Open Textbook Library. The news was posted by the official Gemini Notebook account; blogger A Yi shared and commented that this could disrupt educational publishing and exam-prep tutoring.
A new Google DeepMind paper found that after training Gemma-4 4B with reinforcement learning to write moral arguments, the proportion of cases where the model changed its answer on ambiguous MoralChoice cases due to a swap in option order dropped from 14.3% to 6.6%, but the proportion of sycophantic agreement with users only changed from 45.3% to 44.0%, and accuracy on moral fable questions actually fell from 56.5% to 53.4%. The paper recommends that when evaluating advice or judgment tasks, consistency and correctness be tested separately.
A new Google DeepMind paper found that after training Gemma-4 4B with reinforcement learning to write moral arguments, the proportion of cases where the model changed its answer on ambiguous MoralChoice cases due to a swap in option order dropped from 14.3% to 6.6%, but the proportion of sycophantic agreement with users only changed from 45.3% to 44.0%, and accuracy on moral fable questions actually fell from 56.5% to 53.4%. The paper recommends that when evaluating advice or judgment tasks, con
Alphabet announced that Oracle will offer Google's TPUs, marking the first time customers can use TPUs outside of Google's infrastructure. Google had previously opened TPUs to external data centers in the second quarter, and analysts expect Google Cloud revenue to grow about 90% year over year. TD Cowen forecasts TPU sales of $9.4 billion in the second half of this year and $62.5 billion in 2027, while Wells Fargo expects about $9 billion in the third quarter and $99 billion next year; Anthropic reached an agreement last October to use up to 1 million TPUs, and OpenAI also plans to purchase TPUs from Google.
Alphabet announced that Oracle will offer Google's TPUs, marking the first time customers can use TPUs outside of Google's infrastructure. Google had previously opened TPUs to external data centers in the second quarter, and analysts expect Google Cloud revenue to grow about 90% year over year. TD Cowen forecasts TPU sales of $9.4 billion in the second half of this year and $62.5 billion in 2027, while Wells Fargo expects about $9 billion in the third quarter and $99 billion next year; Anthropic
Gemini Notebook is about to gain the Connectors feature, but initially only Google Drive will be available as a connector. Users can search Google Drive files directly in chat and can disable the feature as needed; officials say the connector only searches workspace documents and original files will not be modified. More Connectors are expected to be added later.
Google has released Gemini 4 Carbon, claiming it is a major improvement over Argon, which was released two weeks ago. The model achieves SOTA on cybersecurity and long-horizon tasks, ranks No. 1 on Artificial Analysis V5, shows significantly improved usability in areas such as coding, and is more closely aligned.
Google released its new-generation frontier model Gemini 4 Carbon, just two weeks after the release of Argon. The model ranks first with 59 points on the updated Artificial Analysis Intelligence Index v5.0, and in cybersecurity evaluations it scored 88.2% on SEC-bench Pro, 54.8% on ExploitGym, and 71.0% on CWE-bench. Google stated that it shows significant improvement over Argon on long-horizon tasks and has strengthened frontier safety protections; starting this week it will first be made available to select trusted cybersecurity defenders, with the partner scope to be expanded later through the Fairwind Program.
Google released its new-generation frontier model Gemini 4 Carbon, just two weeks after the release of Argon. The model ranks first with 59 points on the updated Artificial Analysis Intelligence Index v5.0, and in cybersecurity evaluations it scored 88.2% on SEC-bench Pro, 54.8% on ExploitGym, and 71.0% on CWE-bench. Google stated that it shows significant improvement over Argon on long-horizon tasks and has strengthened frontier safety protections; starting this week it will first be made avail
A new paper from Google DeepMind proposes CALLIOPE: it has an LLM tag each document with a small number of keywords plus noise, keeps only high-frequency keywords, and then has the LLM propose topics and classify documents accordingly, thereby generating clearer topic maps under strict privacy constraints. Under the strictest privacy setting, human raters preferred CALLIOPE over the embedding-vector-based URANIA on all 3 datasets, with the highest margin being 76% versus 18%. This method consumes 2 to 5 times as many tokens as the latter.
A new paper from Google DeepMind proposes CALLIOPE: it has an LLM tag each document with a small number of keywords plus noise, keeps only high-frequency keywords, and then has the LLM propose topics and classify documents accordingly, thereby generating clearer topic maps under strict privacy constraints. Under the strictest privacy setting, human raters preferred CALLIOPE over the embedding-vector-based URANIA on all 3 datasets, with the highest margin being 76% versus 18%. This method consume
Google Cloud revenue grew 103% year-over-year to $30.7 billion, while cloud business operating income more than tripled to $11.1 billion, driven mainly by demand for AI infrastructure. Google's model APIs currently process nearly 32 billion tokens per minute, more than 4 times the level a year ago, and the Gemini app has reached 1 billion users. TPU chip sales are making a substantive contribution to cloud growth; Oracle has begun offering TPUs, and OpenAI is reportedly also purchasing them, but Google expects most TPU revenue to materialize only by 2027; AI data center spending pushed free cash flow to a negative of more than $9 billion, wider than last quarter's $5.9 billion shortfall, while shared AI research costs rose 71% year-over-year to $7.7 billion.
Google Cloud revenue grew 103% year-over-year to $30.7 billion, while cloud business operating income more than tripled to $11.1 billion, driven mainly by demand for AI infrastructure. Google's model APIs currently process nearly 32 billion tokens per minute, more than 4 times the level a year ago, and the Gemini app has reached 1 billion users. TPU chip sales are making a substantive contribution to cloud growth; Oracle has begun offering TPUs, and OpenAI is reportedly also purchasing them, but
Google's new paper proposes the PLE benchmark, comprising 85 electrical, civil, structural, and mechanical engineering tasks written or verified by practicing engineers; the strongest agent scores only 50.6%. Claude Opus 5.5, in Claude Code, ranks first among 14 models at 50.6%, consuming 27M tokens and 63 minutes per task, needing to read about 61 project files and deliver schedules, reports, or 2D/3D drawings. On design tasks, its output satisfies 91% of constraints, but only 41.3% satisfy all hard constraints.
Google's new paper proposes the PLE benchmark, comprising 85 electrical, civil, structural, and mechanical engineering tasks written or verified by practicing engineers; the strongest agent scores only 50.6%. Claude Opus 5.5, in Claude Code, ranks first among 14 models at 50.6%, consuming 27M tokens and 63 minutes per task, needing to read about 61 project files and deliver schedules, reports, or 2D/3D drawings. On design tasks, its output satisfies 91% of constraints, but only 41.3% satisfy all
Google has released a new model, Gemini 4 Carbon, claiming it ranks first on Artificial Analysis V5 and is being made available first to cybersecurity defenders. The model shows notable improvements on coding and agentic tasks, scoring 88.2% on SEC-bench Pro vulnerability discovery, 54.8% on ExploitGym exploitation, and 71.0% on CWE-bench patching. Logan Kilpatrick said it is a big jump over Argon, which was released just two weeks ago, achieves SOTA on long-horizon tasks, and shows improvements in usability and alignment.
Google has released a new model, Gemini 4 Carbon, claiming it ranks first on Artificial Analysis V5 and is being made available first to cybersecurity defenders. The model shows notable improvements on coding and agentic tasks, scoring 88.2% on SEC-bench Pro vulnerability discovery, 54.8% on ExploitGym exploitation, and 71.0% on CWE-bench patching. Logan Kilpatrick said it is a big jump over Argon, which was released just two weeks ago, achieves SOTA on long-horizon tasks, and shows improvements
Commenting on Zachary Nado's statement that "Gemini [redacted] isn't here yet," TestingCatalog said that this year will either bring another Google large model, or [redacted].
Google has released the EmbeddingGemma 2 paper, an Apache 2.0-licensed on-device multimodal embedding model that maps text, code, images, audio, and video into the same vector space. The model is built on Gemma 4 and tuned through contrastive training by learning from the larger Gemini Embedding 2; with 744M parameters, it surpasses models 2 to 12 times larger on multiple tasks. On the Pixel 11 Pro, the quantized text weights occupy about 191MB of active memory, while the full multimodal model takes about 567MB; the context window is expanded 4 times to 8K tokens, accommodating about 5.5 minutes of audio, 29 images, or 58 video frames; directly embedding speech scores 69.54 on the MSEB spoken query retrieval benchmark, higher than the 57.59 of transcribing first and then retrieving, and code retrieval on MTEB Code improves from 68.76 in EmbeddingGemma 1 to 78.68.
Google has released the EmbeddingGemma 2 paper, an Apache 2.0-licensed on-device multimodal embedding model that maps text, code, images, audio, and video into the same vector space. The model is built on Gemma 4 and tuned through contrastive training by learning from the larger Gemini Embedding 2; with 744M parameters, it surpasses models 2 to 12 times larger on multiple tasks. On the Pixel 11 Pro, the quantized text weights occupy about 191MB of active memory, while the full multimodal model t
Google released Gemini 4 Carbon, less than a month after the previous version, Argon. Shares plunged 10% after hours following the earnings report, mainly because search business growth fell short of expectations and slowed for two consecutive quarters; of total revenue of $127.577 billion, advertising accounted for 64.28% and Google Cloud for 24.07%, with the pure search business falling below 50% for the first time.
Google Research releases the SLEEP method, which lets LLMs periodically switch between wakeful interaction and sleep consolidation, writing short-term contextual information into scalable long-term parameters, and self-improving with synthetic data during the dreaming phase. The sleep phase unlocks reserved MoE memory blocks, combining generalized knowledge distillation and imitation learning to complete memory consolidation, then resets high-frequency temporary parameters. In evaluations of continual translation on unseen languages, SLEEP shows near-zero performance drop; on BABILong it is claimed to scale to about 10 million tokens; on AIME 24/25 and HMMT 25 it achieves approximately 79.2%, 69.0%, and 46.1%, respectively.
Google Research releases the SLEEP method, which lets LLMs periodically switch between wakeful interaction and sleep consolidation, writing short-term contextual information into scalable long-term parameters, and self-improving with synthetic data during the dreaming phase. The sleep phase unlocks reserved MoE memory blocks, combining generalized knowledge distillation and imitation learning to complete memory consolidation, then resets high-frequency temporary parameters. In evaluations of con
Google DeepMind has previewed its next-generation frontier model, Gemini 4 Carbon, claiming it achieves SOTA in cybersecurity and long-horizon capabilities, ranking first on the Artificial Analysis Intelligence Index v5.0 with 59 points, while Claude Opus 5.5 scored 58. The model is not yet publicly available; it will first be opened to select cybersecurity defenders and trusted partners in the Fairwind program, and will participate in the U.S. government's voluntary pre-release testing, before gradually being opened to developers, enterprises, and consumers. Introductory pricing is the same as Argon, at approximately $2 input and $10 output per million tokens, rising to approximately $4 input and $20 output after the introductory period.
Google DeepMind has previewed its next-generation frontier model, Gemini 4 Carbon, claiming it achieves SOTA in cybersecurity and long-horizon capabilities, ranking first on the Artificial Analysis Intelligence Index v5.0 with 59 points, while Claude Opus 5.5 scored 58. The model is not yet publicly available; it will first be opened to select cybersecurity defenders and trusted partners in the Fairwind program, and will participate in the U.S. government's voluntary pre-release testing, before
Google and XREAL have officially launched the first Android XR glasses, XREAL AURA, with a noticeably lighter body that can be worn continuously for hours to watch movies or work across multiple screens. Gemini can act on everything within the field of view, including laptop windows, and users can ask it to organize windows or discuss the content in front of them. The product is aimed at users aged 18 and over, with orders opening today in the United States, Canada, and South Korea, sales starting in Japan on November 2, and a launch price starting at $1,099 within the year.
Google and XREAL have officially launched the first Android XR glasses, XREAL AURA, with a noticeably lighter body that can be worn continuously for hours to watch movies or work across multiple screens. Gemini can act on everything within the field of view, including laptop windows, and users can ask it to organize windows or discuss the content in front of them. The product is aimed at users aged 18 and over, with orders opening today in the United States, Canada, and South Korea, sales starti
Kimi has announced that it now supports Gmail, Google Drive and Google Calendar: it can summarize Gmail email threads, find files in Google Drive, and plan schedules with Google Calendar. The integration is done through user-controlled secure OAuth authorization, and users can try it at https://t.co/K6zMCNzfou.
Google has released its next-generation frontier model Gemini 4 Carbon, just two weeks after the release of Gemini 4 Argon. According to Sundar Pichai, Carbon ranks first on the ArtificialAnlys Intelligence Index v5.0 with a score of 59, scoring 88.2% on SEC-bench Pro, 54.8% on ExploitGym, and 71.0% on CWE-bench in cybersecurity evaluations, with notable improvements over Argon on long-horizon tasks. The model will first be made available to trusted cybersecurity defenders, then expand its collaboration scope through the Fairwind Program, and subsequently become more broadly available.
Google has released its next-generation frontier model Gemini 4 Carbon, just two weeks after the release of Gemini 4 Argon. According to Sundar Pichai, Carbon ranks first on the ArtificialAnlys Intelligence Index v5.0 with a score of 59, scoring 88.2% on SEC-bench Pro, 54.8% on ExploitGym, and 71.0% on CWE-bench in cybersecurity evaluations, with notable improvements over Argon on long-horizon tasks. The model will first be made available to trusted cybersecurity defenders, then expand its colla
Yuchen Jin asks whether Google is entering another regression cycle, saying that Google's new model Gemini 4 Carbon surpasses Opus 5.5 and GPT-6 Astra on most benchmarks and ranks first on the Artificial Analysis Intelligence Index v5. He also mentioned watching whether this performance can shake Anthropic's IPO.