David Rojas

Day 25/ We are doubling all usage limits for all Pro 100/200/500 tiers and Business Premium, effective today. We heard your kind feedback and won’t be reducing usage limits for Pro 200, so your Work and Codex usage will stay the same. We’re also going to continue offering extended limits for dots for now, and usage of your dot will still continue to be separate from your default usage. And we'll of course also keep shipping better models and other efficiency improvements that increase your effective usage over time.

Reachy Mini has wheels now! Our little open-source AI robot can finally explore beyond the desktop. And that's not all: 📱 New mobile app: no computer needed 👑 NFC hats that unlock new personalities and apps Physical AI just got a lot more fun. Watch it👇

Full Mario 64 in Mixed Reality! Running in real 3D (no stereoscopic crap) natively at 90FPS and also running on my desk! 😋 Again, Opus 5.5 + PICO CLI combo made it quite straightforward. Such a bliss to play an oldie at 90 FPS in 3D. Disclaimer: I own an original physical copy of Mario 64. This project is not associated with Nintendo or PICO.

🧊 GIVEAWAY 🧊 Win Coffee for Life + a Cumulus machine. 10 more winners get a machine. Find Cumulus at Target, post a photo with #SomethingColdIsHere, tag + follow @cumuluscoffeeco. Ends 10/30 12pm ET.

Introducing Fish Audio Realtime STT (preview): speech-to-text that understands the conversation, not just the words. It knows when to answer, when to keep talking, and when to stop. Ranked #1 of 20 on Cekura’s STT benchmark (2.21% overall word error rate). Built for voice agents. Try it now 👇

FYI you can get enrolled into the Original Content Reward Program without meeting the criteria. If you share Grok Bot templates you can be enrolled into the Grok Bot Creator Program which auto enrolls you into the Original Content Reward since payouts are currently through that program.

While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected. Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions. Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent. But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models? We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem. Read the full report: cohere.com/blog/concerning-b…

As a builder in the AI video space, I was terrified about how much markup API providers charge for simply routing the models. Higgsfield API adds a 40% markup even with its maximum advertised discount. Fal adds 100%. Runway adds 414% (Seedance 2 480p) I might be hated by the industry, but : 
 Today, we’re launching Sume API, the first image & video model provider with 0% markup. For example with Seedance 2.5 1080p • 50% cheaper than fal • 50% cheaper than higgsfield api (29% even with biggest discount) • 28.02% cheaper than kie ai The routing infrastructure layer is never the moat. We believe that APIs should be commoditized, and the moats will be built at the application layer. Check the pricing comparisons below, and see how the broken industry is saved.

“Just my rifle, my pony, and me.“

defining a new category: phone-use an mcp/api that gives ai agents their own phones cloud phones. real androids. real iphones. tell your agent what to do, and it can operate a phone just like you do. today, ai agents face three fundamental problems and phoneuse solves them 🧵

php2shell Tomorrow is a helluva day.

Unitree Introducing: H2-X Kung Fu Legend Guardian🤖 Powerful performance to serve and guard you, with a maximum endurance of approximately 8 hours. Please use the robot safely and in a friendly manner.

A new speech-to-text model of only 1.06MB can transcribe audio ~436× faster than real time on ONE CPU thread! No GPU. AuroraASR-1M is a super super small local speech-recognition model with these stats 🧠 949K parameters 💾 FP32 → 3.81MB 💾 FP16 → 1.91MB 💾 INT8 → 1.06MB 🔓 Apache 2.0 🖥️ CPU or GPU The developer measured the full end-to-end pipeline — feature extraction + speech recognition + CTC decoding at ... 🖥️ 1 CPU thread → ~436× real time 🔥 RTX PRO 6000 Blackwell → ~1,362× real time 🎯So 10 minutes of audio could be processed in about 1.4 seconds at the reported single-thread CPU rate. So to underscore, this isn't using an external language model! It's a tiny Conformer-CTC model trained on LibriSpeech and can run through Transformers locally. There's a tradeoff, accuracy AuroraASR-1M test-clean WER → 11.43% Whisper Tiny English → 5.6% So Whisper Tiny is substantially more accurate. Aurora is also 🇺🇸 English only 🔡 lowercase output ❌ no punctuation A 1MB model can now turn speech into text locally at hundreds of times real time on a single CPU core. That makes completely offline transcription, voice commands and embedded voice features almost absurdly cheap. 🔗 Hugging Face model in ALT.