GPT-6 Astra Tops Terminal Bench 4.0 at Half the Cost
OpenAI's GPT-6 Astra has reached #1 on Terminal Bench 4.0 using the Codex harness, reportedly at 50% of the cost of the runner-up.
OpenAI's GPT-6 Astra has reached #1 on Terminal Bench 4.0 using the Codex harness, reportedly at 50% of the cost of the runner-up.
Resemble AI is joining Summer Signal '26 as GMI Cloud's multimodal detection partner, covering synthetic speech, voice cloning, TTS, watermarking, and detection…
Perplexity details new serving infrastructure (Ivy, Tulip, ROSE) that improves throughput and lowers latency and cost for online and batch embedding workloads.
Perplexity explains how Ivy, its HTTP gateway, handles request parsing, tokenization, templating, and batch splitting before sending workloads to Tulip inferenc…
Perplexity announced the publication of research on how it built state-of-the-art serving infrastructure for the embedding and ranking models behind every answe…
OpenAI's GPT-6-Astra is now available on Requesty, with input at $10/M, cached input at $1/M, and output at $50/M.
Command Code announced that GPT-6 Astra is now live on its platform.
Grok's secure financial account connection via Plaid is now rolling out in Canada for both free and paid users, enabling AI-powered personal finance insights.
OpenAI clarifies Astra rollout details: Plus and Business users who miss access today will get a banked reset by end of day, and newly created accounts by 8pm P…
Grok revealed the winners of its AI video contest, in which users created scenes from Homer's The Odyssey with Grok Imagine for a chance to win $100K.