21 Signals being tracked, weekly summary from the last 7 days:
Site: 3signals - X: @3signalsai
November 14, 2026
Share: X
This is the weekly summary of signals from the last 7 days. The 3 newest signals are first, followed by 18 more in reverse chronological order. Open the full signal list
Weekly summary: 3 new signals first
1. AssemblyAI achieves 70% cost savings on speech-to-text inference using AWS EC2 Spot Instances. (title shortened)
inference-infrastructure - production, business - November 14, 2026
What changed? Results By running Amazon EC2 Spot Instances with GPUs in production, AssemblyAI achieved 70 percent cost savings on inference compute compared to On-Demand pricing. As the platform grew nearly 4x year-over-year in compute demand, the Spot-based architecture kept unit economics sustainable.
From: aws - source
Source context: AssemblyAI achieves 70% cost savings on speech-to-text inference using AWS EC2 Spot Instances with GPU diversification and queue-based retries. Evidence: Results By running Amazon EC2 Spot Instances with GPUs in production, AssemblyAI achieved 70 percent cost savings on inference compute compared to On-Demand pricing. As the platform grew nearly 4x year-over-year in compute demand, the Spot-based architecture kept unit economics sustainable.
Excerpt: Results By running Amazon EC2 Spot Instances with GPUs in production, AssemblyAI achieved 70 percent cost savings on inference compute compared to On-Demand pricing. As the platform grew nearly 4x year-over-year in compute demand, the Spot-based architecture kept unit economics sustainable.
Why is this signal important? This matters because serving improvements can make AI products faster and cheaper to run.
2. NVIDIA NeMo Switchyard optimizes model routing for coding agents on Amazon Bedrock, reducing task costs
agent-workflows, inference-infrastructure - open-source, production - November 14, 2026
What changed? So when you route a coding agent’s requests between models, judge the routing by cost per solved task: the total cost of your runs divided by the number of runs whose tests pass. NVIDIA NeMo Switchyard is an open source library that helps an AI agent choose which model handles each request.
From: aws - source
Source context: NVIDIA NeMo Switchyard optimizes model routing for coding agents on Amazon Bedrock, reducing task costs. Evidence: So when you route a coding agent’s requests between models, judge the routing by cost per solved task: the total cost of your runs divided by the number of runs whose tests pass. NVIDIA NeMo Switchyard is an open source library that helps an AI agent choose which model handles each request.
Excerpt: So when you route a coding agent’s requests between models, judge the routing by cost per solved task: the total cost of your runs divided by the number of runs whose tests pass. [excerpt shortened]
Why is this signal important? This matters because open-source AI tooling is becoming a larger part of production engineering work.
3. GS Caltex accelerates model deployment using Amazon SageMaker Canvas and Pipelines, reducing time from a week to hours
ai-products, inference-infrastructure - production, business - November 14, 2026
What changed? 5. Schedule inference every 30 minutes GS Caltex runs inference every 30 minutes.
From: aws - source
Source context: GS Caltex accelerates model deployment using Amazon SageMaker Canvas and Pipelines, reducing time from a week to hours. Evidence: 5. Schedule inference every 30 minutes GS Caltex runs inference every 30 minutes.
Excerpt: 5. Schedule inference every 30 minutes GS Caltex runs inference every 30 minutes.
Why is this signal important? This matters because serving improvements can make AI products faster and cheaper to run.
4. STABL launches to address AI's transformative potential and societal bottlenecks with a focus on resilience
ai-safety - safety, research - November 14, 2026
What changed? We favor a broad portfolio of defenses to manage AI risks, of which alignment is only one part. Resilience, the most neglected component today, must become central.
From: arvind-narayanan - source
Source context: STABL launches to address AI's transformative potential and societal bottlenecks with a focus on resilience. Evidence: We favor a broad portfolio of defenses to manage AI risks, of which alignment is only one part. Resilience, the most neglected component today, must become central.
Excerpt: We favor a broad portfolio of defenses to manage AI risks, of which alignment is only one part. Resilience, the most neglected component today, must become central.
Why is this signal important? This matters because STABL launches to address AI's transformative potential and societal bottlenecks with a focus on resilience.
5. Amplitude's Wave agent accelerates product improvements using OpenAI models
agent-workflows, ai-products - production, business, open-source - November 14, 2026
What changed? How Amplitude ships product improvements in 1–2 days with OpenAI Built primarily on OpenAI models, Amplitude’s proactive product agent, Wave, identifies what to build, helps teams ship changes with coding agents, and measures the impact.
Article: Amplitude's Wave agent accelerates product improvements using OpenAI models
From: openai - source
Source context: Amplitude's Wave agent accelerates product improvements using OpenAI models. Evidence: How Amplitude ships product improvements in 1–2 days with OpenAI Built primarily on OpenAI models, Amplitude’s proactive product agent, Wave, identifies what to build, helps teams ship changes with coding agents, and measures the impact.
Excerpt: How Amplitude ships product improvements in 1–2 days with OpenAI Built primarily on OpenAI models, Amplitude’s proactive product agent, Wave, identifies what to build, helps teams ship changes with coding agents, and measures the impact.
Why is this signal important? This matters because open-source AI tooling is becoming a larger part of production engineering work.
6. Roblox CEO David Baszucki sees AI as transforming game creation, enabling anyone to build games. (title shortened)
agent-workflows, ai-products - business, production, open-source - November 14, 2026
What changed? We could be approaching the time where “You Make the Game” really comes true, and where making a game is accessible to everyone. It’s really fun, I’ll tell you a bit about some of the experiments we’ve been having with Roblox Build .
From: ben-thompson - source
Source context: Roblox CEO David Baszucki sees AI as transforming game creation, enabling anyone to build games and positioning Roblox as an aggregator in a world of abundant content. Evidence: We could be approaching the time where “You Make the Game” really comes true, and where making a game is accessible to everyone. It’s really fun, I’ll tell you a bit about some of the experiments we’ve been having with Roblox Build .
Excerpt: We could be approaching the time where “You Make the Game” really comes true, and where making a game is accessible to everyone. It’s really fun, I’ll tell you a bit about some of the experiments we’ve been having with Roblox Build .
Why is this signal important? This matters because open-source AI tooling is becoming a larger part of production engineering work.
7. Jensen Huang's sudden departure as Nvidia CEO marks a pivotal shift for the company and the tech industry
ai-products, ai-safety - business, safety, research - November 14, 2026
What changed? This week’s coverage was dominated, unsurprisingly, by the shocking and sudden end of Jensen Huang’s 32 year tenure as CEO of Nvidia. The end came on Thursday morning, and just a few hours later, we recorded a Sharp Tech episode recounting Huang’s legendary history, figuring out what happened, and looked forward to what this might mean for Nvidia, the tech industry, and America going forward.
From: ben-thompson - source
Source context: Jensen Huang's sudden departure as Nvidia CEO marks a pivotal shift for the company and the tech industry. Evidence: This week’s coverage was dominated, unsurprisingly, by the shocking and sudden end of Jensen Huang’s 32 year tenure as CEO of Nvidia. The end came on Thursday morning, and just a few hours later, we recorded a Sharp Tech episode recounting Huang’s legendary history, figuring out what happened, and looked forward to what this might mean for Nvidia, the tech industry, and America going forward.
Excerpt: The Shocking Demise of Jensen Huang. This week’s coverage was dominated, unsurprisingly, by the shocking and sudden end of Jensen Huang’s 32 year tenure as CEO of Nvidia.
Why is this signal important? This matters because Jensen Huang's sudden departure as Nvidia CEO marks a pivotal shift for the company and the tech industry.
8. AWS launches official emulator for local testing of services without paid API keys
ai-products - release, production, business - November 14, 2026
What changed? aws-emulator aws-emulator Finally, an official emulator from AWS to run tests against a local clone of their many different services without having to provision a paid API key to do so. It's written in Rust by a huge swarm of agents - 29,684 commits in the first three weeks of the project.
Article: AWS launches official emulator for local testing of services without paid API keys
From: simon-willison - source
Source context: AWS launches official emulator for local testing of services without paid API keys. Evidence: aws-emulator aws-emulator Finally, an official emulator from AWS to run tests against a local clone of their many different services without having to provision a paid API key to do so. It's written in Rust by a huge swarm of agents - 29,684 commits in the first three weeks of the project.
Excerpt: aws-emulator aws-emulator Finally, an official emulator from AWS to run tests against a local clone of their many different services without having to provision a paid API key to do so. [excerpt shortened]
Why is this signal important? This matters because AWS launches official emulator for local testing of services without paid API keys.
9. Google launches Gemini 4 Hydrogen as a low-cost model for everyday AI tasks
model-releases, ai-products, evaluations - release, business, research, production - November 14, 2026
What changed? Hydrogen is built to stand on its own as a model for everyday work — or to pair with Carbon and Argon as the high-efficiency execution engine in multi-model workflows. Most notably, Hydrogen is the new free tier model, and is the model that will power AI features in Google Search.
Article: Google launches Gemini 4 Hydrogen as a low-cost model for everyday AI tasks
From: simon-willison - source
Source context: Google launches Gemini 4 Hydrogen as a low-cost model for everyday AI tasks. Evidence: Hydrogen is built to stand on its own as a model for everyday work — or to pair with Carbon and Argon as the high-efficiency execution engine in multi-model workflows. Most notably, Hydrogen is the new free tier model, and is the model that will power AI features in Google Search.
Excerpt: Hydrogen is built to stand on its own as a model for everyday work — or to pair with Carbon and Argon as the high-efficiency execution engine in multi-model workflows. [excerpt shortened]
Why is this signal important? This matters because model capability is shifting what builders can expect from current tools.
10. Datasette-leaflet 0.3 updates referrer policy for OpenStreetMap tile requests
ai-products - release, business, open-source - November 14, 2026
What changed? It now sets the referrer policy on the tile requests themselves to strict-origin-when-cross-origin , passing the host but not the full path to the page. Tags: leaflet , datasette , openstreetmap.
Article: Datasette-leaflet 0.3 updates referrer policy for OpenStreetMap tile requests
From: simon-willison - source
Source context: Datasette-leaflet 0.3 updates referrer policy for OpenStreetMap tile requests. Evidence: It now sets the referrer policy on the tile requests themselves to strict-origin-when-cross-origin , passing the host but not the full path to the page. Tags: leaflet , datasette , openstreetmap
Excerpt: It now sets the referrer policy on the tile requests themselves to strict-origin-when-cross-origin , passing the host but not the full path to the page. Tags: leaflet , datasette , openstreetmap
Why is this signal important? This matters because Datasette-leaflet 0.3 updates referrer policy for OpenStreetMap tile requests.
11. Meta commits to clean manufacturing in data center supply chains with new agreements in cement, steel, and freight
ai-products, ai-safety - business, safety, release, production - November 14, 2026
What changed? Our multi-year commitments give suppliers the revenue certainty they need to build these solutions, and stepping in early helps demonstrate these technologies so the broader market can adopt them. These agreements build on a track record of testing, validating, and scaling lower carbon infrastructure solutions across our data center footprint.
From: mark-zuckerberg - source
Source context: Meta commits to clean manufacturing in data center supply chains with new agreements in cement, steel, and freight. Evidence: Our multi-year commitments give suppliers the revenue certainty they need to build these solutions, and stepping in early helps demonstrate these technologies so the broader market can adopt them. These agreements build on a track record of testing, validating, and scaling lower carbon infrastructure solutions across our data center footprint.
Excerpt: Our multi-year commitments give suppliers the revenue certainty they need to build these solutions, and stepping in early helps demonstrate these technologies so the broader market can adopt them. These agreements build on a track record of testing, validating, and scaling lower carbon infrastructure solutions across our data center footprint.
Why is this signal important? This matters because Meta commits to clean manufacturing in data center supply chains with new agreements in cement, steel, and freight.
12. GPT-6.1 Sol achieves perfect accuracy in benchmarking across 60 synthetic tasks
evaluations - research, production - November 13, 2026
What changed? Run identical tasks through OpenRouter and compare graded accuracy, billed cost, and request time. On our 60 synthetic tasks, GPT-6.1 Sol passed 180/180 trials across three repeats under the tested contracts and settings.
Article: GPT-6.1 Sol achieves perfect accuracy in benchmarking across 60 synthetic tasks
From: openrouter - source
Source context: GPT-6.1 Sol achieves perfect accuracy in benchmarking across 60 synthetic tasks. Evidence: Run identical tasks through OpenRouter and compare graded accuracy, billed cost, and request time. On our 60 synthetic tasks, GPT-6.1 Sol passed 180/180 trials across three repeats under the tested contracts and settings.
Excerpt: Run identical tasks through OpenRouter and compare graded accuracy, billed cost, and request time. On our 60 synthetic tasks, GPT-6.1 Sol passed 180/180 trials across three repeats under the tested contracts and settings.
Why is this signal important? This matters because frontier AI economics and compute needs are scaling quickly.
13. Amazon SageMaker HyperPod introduces Data Capture to improve model accuracy by recording live inference data
ai-products, inference-infrastructure - production, business - November 13, 2026
What changed? Without a record of production inference data, teams can’t detect drift, find failure patterns, or use real interactions to improve the model. Amazon SageMaker HyperPod now provides Data Capture, a built-in capability that records live inference inputs and outputs to Amazon Simple Storage Service (Amazon S3) at three independent tiers, with minimal latency impact and zero impact on model availability.
From: aws - source
Source context: Amazon SageMaker HyperPod introduces Data Capture to improve model accuracy by recording live inference data. Evidence: Without a record of production inference data, teams can’t detect drift, find failure patterns, or use real interactions to improve the model. Amazon SageMaker HyperPod now provides Data Capture, a built-in capability that records live inference inputs and outputs to Amazon Simple Storage Service (Amazon S3) at three independent tiers, with minimal latency impact and zero impact on model availability.
Excerpt: Amazon SageMaker HyperPod now provides Data Capture, a built-in capability that records live inference inputs and outputs to Amazon Simple Storage Service (Amazon S3) at three independent tiers, with minimal latency impact and zero impact on model availability. [excerpt shortened]
Why is this signal important? This matters because serving improvements can make AI products faster and cheaper to run.
14. Microsoft identifies 'Lazarus' backdoor technique exploiting dormant neurons in AI models to execute malicious commands
ai-safety, evaluations, ai-products - safety, research, production, business - November 13, 2026
What changed? Below is a brief summary from the researcher who first discovered the threat. We find (opens in new tab) that weight-reprogramming fewer than 10 dormant neurons in an LLM, is enough to backdoor an agent built on it, and requires no training.
From: microsoft-research - source
Source context: Microsoft identifies 'Lazarus' backdoor technique exploiting dormant neurons in AI models to execute malicious commands. Evidence: Below is a brief summary from the researcher who first discovered the threat. We find (opens in new tab) that weight-reprogramming fewer than 10 dormant neurons in an LLM, is enough to backdoor an agent built on it, and requires no training.
Excerpt: Below is a brief summary from the researcher who first discovered the threat. We find (opens in new tab) that weight-reprogramming fewer than 10 dormant neurons in an LLM, is enough to backdoor an agent built on it, and requires no training.
Why is this signal important? This matters because teams are turning AI agents into repeatable production workflows.
15. Microsoft Research solves the 31-year-old KLS conjecture, advancing understanding of randomized algorithms
ai-products, evaluations, model-releases, ai-safety - research, production, business, release - November 13, 2026
What changed? New research has resolved the conjecture. This removes a major theoretical obstacle to understanding the efficiency of randomized algorithms used across science and computing.
From: microsoft-research - source
Source context: Microsoft Research solves the 31-year-old KLS conjecture, advancing understanding of randomized algorithms. Evidence: New research has resolved the conjecture. This removes a major theoretical obstacle to understanding the efficiency of randomized algorithms used across science and computing.
Excerpt: New research has resolved the conjecture. This removes a major theoretical obstacle to understanding the efficiency of randomized algorithms used across science and computing.
Why is this signal important? This matters because frontier AI economics and compute needs are scaling quickly.
16. Google Search integrates Gemini 4 Hydrogen for enhanced capabilities
ai-products, model-releases, evaluations, inference-infrastructure - business, release, production, research - November 13, 2026
What changed? Google Search gets a major upgrade with Gemini 4 Hydrogen Rotating carousel of cards with text "Gemini 4 Hydrogen" "Now in Google Search" and examples from the Google Search UI.
Article: Google Search integrates Gemini 4 Hydrogen for enhanced capabilities
From: google-ai - source
Source context: Google Search integrates Gemini 4 Hydrogen for enhanced capabilities. Evidence: Google Search gets a major upgrade with Gemini 4 Hydrogen Rotating carousel of cards with text "Gemini 4 Hydrogen" "Now in Google Search" and examples from the Google Search UI
Excerpt: Google Search gets a major upgrade with Gemini 4 Hydrogen Rotating carousel of cards with text "Gemini 4 Hydrogen" "Now in Google Search" and examples from the Google Search UI
Why is this signal important? This matters because Google Search integrates Gemini 4 Hydrogen for enhanced capabilities.
17. Google DeepMind's AlphaGenome 2 improves DNA variant prediction accuracy from 16% to 30%
model-releases - release, research - November 13, 2026
What changed? AG2 takes the same million letters plus a condition, which can be a tissue, a cell type, a cell state, or data from a few dozen of your own cells that it has never seen before, and it predicts what the DNA does there specifically. It trained on more than 107 million individual cell profiles from 68 human and mouse tissues.
Article: Google DeepMind's AlphaGenome 2 improves DNA variant prediction accuracy from 16% to 30%
From: packy-mccormick - source
Source context: Google DeepMind's AlphaGenome 2 improves DNA variant prediction accuracy from 16% to 30%. Evidence: AG2 takes the same million letters plus a condition, which can be a tissue, a cell type, a cell state, or data from a few dozen of your own cells that it has never seen before, and it predicts what the DNA does there specifically. It trained on more than 107 million individual cell profiles from 68 human and mouse tissues.
Excerpt: AG2 takes the same million letters plus a condition, which can be a tissue, a cell type, a cell state, or data from a few dozen of your own cells that it has never seen before, and it predicts what the DNA does there specifically. [excerpt shortened]
Why is this signal important? This matters because model capability is shifting what builders can expect from current tools.
18. Jensen Huang resigns from Nvidia, leaving a leadership gap amid its dominance in the AI chip market
ai-safety, ai-products - safety, research, business - November 13, 2026
What changed? They account for 85% of the market . The company has provided financial guarantees to the constellation of upstart AI infrastructure companies, the so-called neoclouds, so they can get their business moving. The signal is supported by 2 sources, including azeem-azhar, casey-newton.
Article: Jensen Huang resigns from Nvidia, leaving a leadership gap amid its dominance in the AI chip market
From: azeem-azhar - source
Source context: Jensen Huang resigns from Nvidia, leaving a leadership gap amid its dominance in the AI chip market. Evidence: They account for 85% of the market . The company has provided financial guarantees to the constellation of upstart AI infrastructure companies, the so-called neoclouds, so they can get their business moving.
Excerpt: They account for 85% of the market . The company has provided financial guarantees to the constellation of upstart AI infrastructure companies, the so-called neoclouds, so they can get their business moving.
Article: Jensen Huang resigns as Nvidia CEO amid scandal, highlighting risks of founder-centric governance
From: casey-newton - source
Source context: Jensen Huang resigns as Nvidia CEO amid scandal, highlighting risks of founder-centric governance. Evidence: Jensen Huang and the end of the ‘lol nothing matters’ era of American politics This column touches on aspects of the AI industry. My fiancé works at Anthropic. See my full ethics disclosure here . On Thursday morning, Nvidia announced that CEO Jensen Huang had resigned from the company. Colette Kress, Nvidia’s longtime chief financial officer, will serve as CEO while the board searches for a permanent replacement. [excerpt shortened]
Excerpt: Jensen Huang and the end of the ‘lol nothing matters’ era of American politics This column touches on aspects of the AI industry. My fiancé works at Anthropic. See my full ethics disclosure here . On Thursday morning, Nvidia announced that CEO Jensen Huang had resigned from the company. [excerpt shortened]
Why is this signal important? This matters because Jensen Huang resigns from Nvidia, leaving a leadership gap amid its dominance in the AI chip market.
19. Google launches Gemini Decision 1, a multimodal decision model with competitive pricing and performance benchmarks
evaluations, ai-products, model-releases - release, business, research, production - November 13, 2026
What changed? Pricing : $0.10 per 1M input tokens, with $0 billed for output. Google says that is about 5x cheaper than Flash-Lite ( Sundar ).
From: alessio-fanelli - source
Source context: Google launches Gemini Decision 1, a multimodal decision model with competitive pricing and performance benchmarks. Evidence: Pricing : $0.10 per 1M input tokens, with $0 billed for output. Google says that is about 5x cheaper than Flash-Lite ( Sundar ).
Excerpt: Pricing : $0.10 per 1M input tokens, with $0 billed for output. Google says that is about 5x cheaper than Flash-Lite ( Sundar ).
Why is this signal important? This matters because frontier AI economics and compute needs are scaling quickly.
20. ArtCraft aims to revolutionize software with AI-driven open-source alternatives to Adobe and Microsoft tools
ai-products - release, open-source, business - November 13, 2026
What changed? “AutoCraft is a completely autonomous software factory that we're going to open source.” AI open source FTW There’s no doubting the ambition of Brandon Thomas and his ArtCraft company. “We can open source the entire world,” he proclaimed at the end of his presentation.
From: alessio-fanelli - source
Source context: ArtCraft aims to revolutionize software with AI-driven open-source alternatives to Adobe and Microsoft tools. Evidence: “AutoCraft is a completely autonomous software factory that we're going to open source.” AI open source FTW There’s no doubting the ambition of Brandon Thomas and his ArtCraft company. “We can open source the entire world,” he proclaimed at the end of his presentation.
Excerpt: “AutoCraft is a completely autonomous software factory that we're going to open source.” AI open source FTW There’s no doubting the ambition of Brandon Thomas and his ArtCraft company. “We can open source the entire world,” he proclaimed at the end of his presentation.
Why is this signal important? This matters because open-source AI tooling is becoming a larger part of production engineering work.
21. Nvidia CEO Jensen Huang's affair with a Chinese spy raises national security concerns and prompts. (title shortened)
ai-safety - safety, business, research - November 13, 2026
What changed? Jensen Huang had regular phone calls with Trump. Jensen and Alice Huang presumably talked.
From: zvi-mowshowitz - source
Source context: Nvidia CEO Jensen Huang's affair with a Chinese spy raises national security concerns and prompts calls for his resignation. Evidence: Jensen Huang had regular phone calls with Trump. Jensen and Alice Huang presumably talked.
Excerpt: Jensen Huang had regular phone calls with Trump. Jensen and Alice Huang presumably talked.
Why is this signal important? This matters because Nvidia CEO Jensen Huang's affair with a Chinese spy raises national security concerns and prompts calls (shortened).
This week’s developments
inference-infrastructure: 1 developments
agent-workflows: 3 developments
- Nov 12: Roblox CEO David Baszucki sees AI as transforming game creation, enabling anyone to build games. (title shortened)
- Nov 12: Amplitude's Wave agent accelerates product improvements using OpenAI models
- Nov 13: NVIDIA NeMo Switchyard optimizes model routing for coding agents on Amazon Bedrock, reducing task costs
ai-products: 9 developments
- Nov 11: AWS launches official emulator for local testing of services without paid API keys
- Nov 12: Meta commits to clean manufacturing in data center supply chains with new agreements in cement, steel, and freight
- Nov 12: Google Search integrates Gemini 4 Hydrogen for enhanced capabilities
- Nov 12: Microsoft Research solves the 31-year-old KLS conjecture, advancing understanding of randomized algorithms
- Nov 13: ArtCraft aims to revolutionize software with AI-driven open-source alternatives to Adobe and Microsoft tools
- Nov 13: Jensen Huang's sudden departure as Nvidia CEO marks a pivotal shift for the company and the tech industry
- Nov 13: Amazon SageMaker HyperPod introduces Data Capture to improve model accuracy by recording live inference data
- Nov 13: Datasette-leaflet 0.3 updates referrer policy for OpenStreetMap tile requests
- Nov 13: GS Caltex accelerates model deployment using Amazon SageMaker Canvas and Pipelines, reducing time from a week to hours
ai-safety: 4 developments
- Nov 10: Nvidia CEO Jensen Huang's affair with a Chinese spy raises national security concerns and prompts. (title shortened)
- Nov 10: STABL launches to address AI's transformative potential and societal bottlenecks with a focus on resilience
- Nov 10: Microsoft identifies 'Lazarus' backdoor technique exploiting dormant neurons in AI models to execute malicious commands
- Nov 12: Jensen Huang resigns from Nvidia, leaving a leadership gap amid its dominance in the AI chip market
model-releases: 2 developments
- Nov 12: Google launches Gemini 4 Hydrogen as a low-cost model for everyday AI tasks
- Nov 13: Google DeepMind's AlphaGenome 2 improves DNA variant prediction accuracy from 16% to 30%
evaluations: 2 developments
- Nov 12: GPT-6.1 Sol achieves perfect accuracy in benchmarking across 60 synthetic tasks
- Nov 12: Google launches Gemini Decision 1, a multimodal decision model with competitive pricing and performance benchmarks
What's new with 3signals
Recent product improvements:
- Vibe Check section (2026-06-11): 3signals now has a Vibe Check section for surfacing community-validated momentum alongside the system's curated signal picks. Details
- Interactive wiki graph view (2026-05-18): The 3signals wiki now includes an Obsidian-style graph for exploring how signals connect to topics, concepts, authors, and source evidence. Details
- Front-end and back-end split for faster site delivery (2026-05-17): 3signals now serves the public website from Vercel while Railway keeps running the API, cron jobs, and content generation pipeline. Details
Staged future improvements:
- Fold reader feedback into presentation scoring so useful signals can be resurfaced with better timing.
- Expand archive analytics so opens, votes, site access, and X posts can be compared by issue.
- Continue tightening source QA for headline strength, evidence fit, and source freshness.