OpenAI's 'Ultrafast' Mode Hits 750 Tokens/Sec, Powered by Cerebras Wafer-Scale Chips
GPT-5.6 Sol now runs 14x faster without sacrificing intelligence. Here's why unbundling speed from model size changes the math for agentic workflows.
The day's biggest AI story, explained at length with its sources. Page 8 of 12.
GPT-5.6 Sol now runs 14x faster without sacrificing intelligence. Here's why unbundling speed from model size changes the math for agentic workflows.
As AI agents bloat context windows with hundreds of tool schemas, Okta's new identity-scoped MCP filtering solves the enterprise tool tax.
The Swedish AI startup just raised $400M, doubled its valuation in eight months, and hit a massive $500M ARR. Here is why enterprise IT is fully buying in.
OpenAI just dropped GPT-5.6-Cyber behind its strict Daybreak Red tier. It boasts a 95% completion rate on exploit prompts, but standard Sol still beats it on benchmarks.
With single-pass 30-second generation, native audio co-generation, and up to 50 multimodal references, Seedance 2.5 sets a new bar for AI video.
OpenAI’s new flagship model family crushes Claude Fable 5 on agentic benchmarks, introducing parallel subagents and massive cost-efficiency gains.
Released on August 7, xAI's new image generation model leapfrogs Midjourney and Meta to claim the 2 spot on the Arena, treating precision editing as a first-class capabi…
Thinking Machines Lab just dropped a massive multimodal MoE with a 1-million token context and a controllable "thinking effort" dial—and it's completely open-source.
In a display of regulatory capture, the two fiercest rivals in AI have found common ground: lobbying the US government to ban open models under the guise of national sec…
As users on the premium Pro 20X plan face multi-day outages for GPT-5.6 Sol Max, a severe bug in the Codex Desktop app's Computer Use service is quietly bricking macOS m…
By ditching the heavy LLM bottleneck, researchers achieved a 97.7% success rate on LIBERO using under 1GB of VRAM on a consumer RTX 4090.
Anthropic's latest flagship model crushes benchmarks and delivers near-Fable intelligence, but early users report alignment quirks and 'Heisenberg' bugs.