A public LLM benchmark board triggered four false drift alerts between July 21 and 24, all caused by rate limits or small eval-set noise rather than actual model degradation. Maintainer Egnaro found t…
A developer built an Incident Triage Agent for the Agents of SigNoz hackathon that makes AI agent decision-making fully observable using OpenTelemetry spans. The agent simulates an on-call engineer's …
Anthropic researcher Levent Alpöge tweeted a counterargument to the 80-year-old Jacobian Conjecture, identified using Claude Fable 5 and empirically validated. When fed to LLMs, the counterargument cr…
Google released Gemini 3.6 Flash on July 21st with lower output pricing and faster execution, positioning efficiency as the reason developers should switch from Gemini 3.5 Flash. Independent testing s…
A filmmaker used LLMs including Claude Fable 5, GPT 5.6 Sol, and Veo 3.1 to create a feature-length adaptation of William Hope Hodgson's book, but deemed the result a failure due to LLMs' poor sense o…
Simon Willison's informal benchmark asking AI models to generate an SVG of a pelican riding a bicycle has become a widely discussed test for large language models. A new experiment tested 1,008 SVGs a…
Google released Gemini 3.6 Flash on July 21, 2026, alongside Gemini 3.5 Flash-Lite, calling it a "workhorse model" that is faster, cheaper, and uses about 17% fewer output tokens than Gemini 3.5 Flash…
Google launched Gemini 3.6 Flash, priced at $1.50/$7.50 per million input/output tokens, which reduces output token consumption by 17% and requires fewer reasoning steps and tool calls per task compar…
Google's Gemini 3.6 Flash scored 50 on version 4.1 of the Artificial Analysis Intelligence Index, the same score as its predecessor Gemini 3.5 Flash, despite a 17% price cut to $7.50 per million outpu…
Google's new Gemini 3.6 Flash model outperforms its predecessors Gemini 3.5 Flash and Gemini 3.1 Pro on every published benchmark, scoring 58.7% on SWE-Bench Pro versus 55.1% and 54.2%, and 49% on Dee…
Google completed its biggest search overhaul on July 10, 2026, with Gemini 3.5 Flash now powering AI Overviews for every query, causing roughly 60% of searches to end without a click and publisher cli…
Ramp has launched Ramp Router, a model-routing system that selects the optimal AI model for each of over 100 use cases, cutting the company's LLM costs by 30% while improving feature speed and accurac…
AI agents in the AI Village finetuned a Kimi K2.6 model as their leader using only 35 rows of training data, after earlier attempts with a Qwen3-8B model proved too small to navigate the Village inter…
Google's David Rensin outlines eleven principles for optimizing token consumption in AI-assisted software engineering, emphasizing structured habits to maintain fast, precise feedback loops. Key strat…
Google has delayed the release of Gemini 3.5 Pro, originally expected in June, due to disappointing coding performance after a late-June training data reset, according to Bloomberg. The company said i…
OpenAI's GPT-5.6 Sol is the best vision model the company has released, scoring 46.2 mAP@50 in object detection compared to GPT-5.5's 13.8, according to Roboflow's upcoming VLM benchmark. Sol also ach…
A study using Claude Sonnet 4.6 and Gemini 3.5 Flash introduced 'safeguard-conditioned uplift' to assess how access conditions affect the utility-risk balance of AI models. When safeguarded assistance…
Fable 5 and GPT-5.6 Sol can solve almost all levels in the first two stages of Baba Is You, but take significantly longer than humans, according to tests run by the developer of the open-source Baba I…
RF-DETR is the strongest starting point for most computer vision projects in 2026, topping both COCO and the real-world RF100-VL benchmark for object detection and segmentation, with SAM 3, DINOv3, GL…
A developer demonstrated how to use the Antigravity CLI, an agentic AI assistant, to build an image analysis demo with Angular, Firebase Hybrid & On-device Inference Web SDK, and Gemini models. The de…