Loading activity…
@marktenenholtz
Member of Technical Staff @perplexity_ai
AI
His early-access testing favored one agent for coding projects, but clear goals and human oversight still mattered.
AI
Perplexity’s Teammates can monitor and manage market orders for users, Arav Srinivas says.
AI
Prime Intellect says revisions to AutomationBench’s tool descriptions, tasks and graders drastically changed model scores.
AI
The reported MATH-500 score rose from 7.9% to 48.1% with the same sampling settings.
AI
Particle says Particle-1 uses roughly 45% fewer tokens on average than DeepSeek V4.1 Flash across varied tasks.
AI
Perplexity says JOLT trains one model as teacher and student, aiming to make the teacher's feedback more useful.
AI
A post quotes Redpoint's Logan Bartlett saying VC incentives can favor raising and deploying more money during a bubble.
AI
Perplexity says the model and agent run on-device, while its separate Hybrid mode can use cloud models.
AI
The user says the evaluation was for a personal project and that decision models have a long way to go.
AI
A quote post says the task is to forecast crypto coins with a holdout set from 2024.
AI
Artificial Analysis says Exa's auto mode with Dynamic Highlights scored 80 on its Search Index, tying Perplexity Search (medium).
AI
One post argues raw-data brokerage will commoditize, while refined post-training tasks will grow more differentiated.
AI
OpenAI Devs says the dot investigated the issue and started a Codex thread he could steer from his phone.
XAI
Perplexity says Pro and Max subscribers have access; a separate post says the model is available to all users.
AI
Perplexity's README says SPACE uses isolated Firecracker microVMs and an HTTP API to run commands and collect output.
AI
A user says a student from Waterloo sent a templated AI-generated response after the user replied to an advice request.
AI
A user says they're playing around with the model and links to a photo.
AI
A post sharing a conversation with Amplitude, Deloitte and Rakuten leaders says the model scored 90%+ on data visualization evaluations.
AI
A user praises what people are building with EmbeddingGemma 2 and welcomes the attention embedding models are getting.
AI
A post argues that human tasks are now too easy to use as AI benchmarks in many domains.
