Vannarot Roeung
Recent posts by Vannarot Roeung
Jev Ultrafast: 7-Second Browser Agent at $0.0039
browser-use shipped an MIT-licensed browser agent that searches Google Flights in 7.073 seconds. Median browser protocol calls fell 1,092 to 101, a 90.8% cut, but wall-clock time fell only 25%. Here is where the seconds and the $0.0039 actually go.
GitHub's Copilot Rust Port Cost $120,000 in Tokens
GitHub published the full telemetry for rewriting the Copilot agent runtime into 832,378 lines of Rust with AI agents: $120,000 in tokens, a 96.22% prompt cache hit rate, and 28 hours of reading for every hour of editing.
Claude Docs and Slides Land as Cowork Merges Into Chat
Anthropic merged Claude Cowork into Claude chat on September 16, 2026, closing an eight-month split between its conversational assistant and its agentic workspace, and added Claude Docs and Claude Slides as betas.
Firefox Smart Window Runs Mistral Small 4 in the Cloud
Mozilla and Mistral put an Apache 2.0 model into Firefox Smart Window. It does not run on your machine, and the smallest published build of Mistral Small 4 is 54.4 GB. The privacy guarantee here is contractual, not architectural.
Open Source Lip Sync: Which Models You Can Ship
The open-source lip sync model most tutorials still recommend first is the one you are least allowed to use. Five models compared on the only axis that decides whether you can ship the output: what the license actually says.
TypeSafe Jev: 0% Hallucination, 67.8% Accurate
TypeSafe AI's Jev ships with two headline numbers that sound contradictory: a 0% hallucination rate and 67.8% accuracy. Both are true at the same time, and the distance between them is the entire product.
Meta One Pricing: 8 Tiers, No Disclosed AI Limits
Meta launched Meta One globally on 15 September 2026, putting AI image generation, AI video and Instagram Restyle behind eight tiers from $2.99 to $499 a month. The announcement names more than 50 features and not one generation limit.
Cloudflare Disallow AI Training: What It Blocks
On 15 September 2026 Cloudflare turned on Disallow AI Training for every customer on every plan. It tells Google, Apple and Microsoft not to train on your pages while leaving their search crawlers free to index you normally. Until now those were the same switch.
Vidu S2: 720p Real-Time Video With a 25 FPS Floor
ShengShu shipped Vidu S2 on 15 September: real-time interactive video at 720p, up from 540p. The paper says it outperforms all baselines, and it does. The margin runs from 1.9% to roughly half, depending on which task you point it at.
AI Model Retirement: How Much Notice You Actually Get
When Suno shipped v6, every older model left the picker the same day. Amazon Bedrock gives most models six months. The length of the warning you get is set by the platform you call the model through, not by the model.
Cline Desktop Runs Open-Weight Agents Outside Your Editor
Cline shipped a desktop app on September 14, 2026, and the interesting part is not the window. Every headline feature was already shipping as an installable package four months ago, when Cline released its SDK.
Qwen3-TTS Ranks #2 and #28 on the Same Leaderboard
Nari Labs says it leads Coval's voice AI benchmarks. The live board shows something more useful: three companies serve the same Qwen3-TTS 1.7B weights and land at 2nd, 4th and 28th of 28, a 10.8x spread in time-to-first-audio.
Real-SWE: Coding Agents Score 38.8% on Private Code
Specific Labs published Real-SWE on September 12, 2026, a coding-agent benchmark built entirely from private production codebases. The top score is 38.8%, from Claude Fable 5.1 in Claude Code. The same generation posts 95% on SWE-bench Verified.
Agnes 3.0 Flash: The Open Weights Aren't the Leaderboard
Agnes AI published Agnes-3.0-Flash Preview, a 33B hybrid-attention multimodal model with a 262,144-token context window, under Apache 2.0 on 11 September 2026. The downloadable weights are a different checkpoint from the Agnes 3.0 Flash listed on Artificial Analysis.
ElevenLabs Music v2.5 Makes Lossless Free on Every Plan
ElevenLabs released Music v2.5 on 11 September 2026 and made one decision that matters more than the model: lossless downloads on every plan, including Free, which gets five a day. Two days after Suno shipped v6, the AI music fight has moved from output quality to export rights.
Gemini for Windows Takes Alt+Space From ChatGPT
Google released the Gemini desktop app for Windows on 10 September 2026, and it opens with Alt + Space. That is the same global hotkey OpenAI's ChatGPT Windows app has used for its companion window since launch.
Cognition SWE-2 vs Fable 5.1: Where It Wins and Loses
Cognition released SWE-2 on 10 September 2026 and one number travelled: 50.0% on FrontierCode 1.1 Main, within a point of Fable 5.1, at 64% lower cost. Three rows below it, in the same table, is a number that did not.
OpenAI Agents API vs Claude Managed Agents
OpenAI released the Agents API in public beta on 10 September 2026, a managed Codex harness that runs the agent loop for you. Anthropic shipped the same shape five months earlier.
Workflow1111 Rebuilds AUTOMATIC1111 as MCP Tools
The Gradio team rebuilt most of AUTOMATIC1111 as a 73-node canvas with nine REST endpoints that double as MCP tools. The project it replaces has not shipped a tagged release since February 2025.
Cursor Projects Puts a Manager Agent Over Your Agents
Cursor shipped Projects in beta on 10 September 2026: a coordinator agent that does not write code, plans the work, and delegates it to subagents on a cloud machine. The launch numbers are real, and neither one means what a headline will say.
MultiMatte Turns Background Removal Into a Prompt
Feyn released MultiMatte on 10 September 2026, a background removal model you steer by naming the subject you want to keep. It returns an alpha matte rather than a binary mask, and its benchmarks come with one caveat worth reading first.
Cohere North Small Translate Beats DeepL, With a Catch
Cohere released North Small Translate 1.0 on 10 September 2026, a 218B model scoring 83.60 on WMT26 against DeepL NextGen at 81.37. The weights are CC BY-NC 4.0, so you cannot use them in anything you sell.
DeepSeek V4.1-Flash Retires V4-Pro and Cuts Prices
DeepSeek shipped V4.1-Flash on 10 September: 552B multimodal MoE, MIT open weights, cache-hit prices down 57%, and every V4-Pro API call rerouted to it from 14 September.
GPT-Live-1 API: Voice Agents Now Bill by the Minute
OpenAI put GPT-Live-1 into the API on 10 September 2026 at $0.05 per minute, billed per second. The price got the headlines. The bigger change is that rate limits are now quoted in concurrent sessions, and the model runs on one endpoint that your existing integration cannot reach.