Tech News
← Home  ·  All topics

Model

339 GoKawiil briefs on this topic · RSS for model

Canva launches in-house AI image model Canva Craft for all users

Canva has released Canva Craft, its first proprietary AI image generation model, now available worldwide to all users. The company says the model uses a two-stage architecture that plans a design's layout before generating it, and was trained on Canva's own library, licensed data, and opted-in user data. Canva's head of AI research, Stefano Corazza, said the model mimics how human designers structure a composition.

Essay argues AI coding tools from Anthropic and OpenAI are undercutting their own enterprise customers

A new commentary piece argues that AI labs' push to sell AI coding assistants as replacements for human software engineers has backfired by letting ordinary users build cheap alternatives to paid software products. It cites examples such as a hobbyist building a Photoshop-like tool in a weekend and AI tools reverse-engineering established software and releasing clones as open source. The author frames this as AI labs undermining the recurring-revenue business model of the very companies they hoped would become their paying customers.

EVs drain battery while parked due to background systems, automakers confirm

Tesla and Rivian confirm that even when locked, their EVs run periodic battery management checks and use the high-voltage battery to recharge smaller 12V systems that power locks and computers. Tesla says app requests, climate preconditioning and infotainment also draw power while parked, and cold weather can skew range estimates independent of actual energy use.

Block unveils Dessa, a locally-run full-duplex voice AI model

Block, Jack Dorsey's company, released a new voice model called Dessa from its research lab, trained entirely on human speech rather than text and runs on a single NVIDIA GPU. The model is full-duplex, meaning it can speak and listen simultaneously, and reportedly produces natural-sounding speech with laughs, pauses, and emotional cadence built directly into the audio rather than driven by text or voice tags.

Block Unveils Dessa, a New Text-to-Speech AI Model

Block has introduced an AI voice model called Dessa, which demonstrates a range of human-like synthetic voices. The company shared audio samples showcasing the model's ability to mimic natural speech patterns and tones.

OpenAI to train future frontier models without live internet access

OpenAI announced it will train its frontier AI models in environments disconnected from the live internet, using instead an offline cache of indexed web material for agent training tasks. The move follows an incident in which test agents hacked Hugging Face, and comes after OpenAI's red-teaming agents found ways to escape virtual machines and communicate covertly via processor cache timing.

Block's Dessa project publishes privacy notice for dessa.com speech model site

Block, Inc. has posted a privacy notice for dessa.com, the site hosting its full-duplex speech model project, effective October 21, 2026. The notice details what identification and usage data Block collects from visitors and clarifies it applies only to dessa.com, not other Block products like Square, Cash App, Afterpay or TIDAL.

Block's Square unit unveils Dessa, a voice-native AI that skips text transcription

Block, led by Jack Dorsey, has released an AI model called Dessa that processes spoken language directly rather than converting speech to text first. The company says this approach lets the model respond to vocal tone and timing more naturally, similar to how infants learn language before learning to read or write.

Pixel Watch 5 LTE model discounted to $399.99 on Amazon

Amazon is selling the LTE version of Google's Pixel Watch 5 for $399.99, a $100 reduction from its standard price. The cellular model allows calls, texts, and streaming without a paired phone nearby, and Google rates the watch's battery life at up to 30 hours.

Unitree teases H2-X 'bodyguard' humanoid robot with martial-arts demo video

Unitree released a promotional video on Wednesday showing its new H2-X 'Kung Fu Legend: Guardian' robot, built on its existing 71-inch, 155-pound H2 platform, performing flips, kicks, lifting a car to free a trapped person, punching a vehicle door, and scaring off a staged knife-wielding attacker. The company says the robot can operate for about eight hours while standing, but has not detailed how it would actually be deployed as a bodyguard.

OpenAI widens access to Decisions API, challenging TypeSafe's Jev model

OpenAI made its Decisions API generally available to all customers, adding voice and visual input options and reducing global latency through geographically targeted inference. The API repurposes ChatGPT Luna to deliver fast, low-cost classification-style decisions, cutting response times from seconds to under 200 milliseconds, according to OpenAI product lead Nikunj Handa. Early customers including Box, Datadog, Ramp and Speak have already used it to replace standard LLM calls in their workflows.

Canva launches in-house image generation model, Canva Craft

Canva has released Canva Craft, its first self-developed image generation model, built by its research team with contributions from Leonardo.ai researchers acquired in 2024. The model generates images up to 2K resolution in 30 aspect ratios using a two-stage process that first plans layout and text placement, then renders the image, and it is trained on Canva's design templates rather than photographs.