225 results for “IRST”
NVIDIA’s DSX bet is tokens per megawatt: Lambda reports 24% more HGX B200 throughput at 85% power. Eos + Emerald cut 4 MW to 3 MW for the grid — Flex-pattern proof, not a shipped DSX Flex site.
Shanghai AI Lab open-sourced Intern-S2 with a plug-and-play Memory Decoder for science tasks. Weights are on Hugging Face; closed-model parity claims stay lab-reported only.
Amodei’s pace-the-frontier call won lab backing — then Trump and Nvidia said no slowdown onstage. Monday’s tape split the story: cyber names jumped while chip infrastructure sold off.
Z.ai released GLM-5.3-Flash, a 320B-parameter multimodal model with only 18B active parameters using hybrid sparse-linear attention. The MIT-licensed model approaches Claude Opus 4.8 on coding benchmarks at one-tenth the inference cost of its predecessor.
OpenAI says Perplexity trusts GPT-6 Astra to write communications, edit software, and monitor production with less oversight. The endorsement suggests Astra's computer-use skills are crossing a trust threshold.
September 2026 watch through 14 Sep: Daybreak keys passed, DeepSeek V4.1-Flash live with Pro remap, Astra and Agents API surfaces landed — Gemini 4 and Apple China launch timing still open.
Andon Labs says GPT-6 Astra averaged $15,515 in its Vending-Bench 2 simulation, nearly triple Claude Fable 5.1's result. The test suggests long-horizon discipline, not raw reasoning, is the next agent battleground.
SenseTime fully open-sourced SenseNova U1, a native unified multimodal model that fuses vision and language without a vision encoder or VAE. The 8B-parameter release matches commercial models on infographic generation and adds native 4K output in its U1.5 update.
NVIDIA took a minority stake in data-center site developer Cloverleaf Infrastructure to accelerate AI factory construction across the U.S. The partnership integrates NVIDIA's DSX platform early in site design to maximize compute output per megawatt.
Amazon Science released SOP-Bench, an open benchmark with 2,000+ tasks from real business SOPs across twelve domains. The framework pairs authentic procedures with working tools and ground-truth answers so teams can validate agents before production deployment.
OpenAI released GPT-Live-1 in its API on September 10, 2026, a full-duplex voice model that listens and speaks simultaneously. OpenAI reports it cuts interruptions by nearly 80% in early tests with Speak.
NVIDIA launched WARDOGS early access and Valheim 1.0 on GeForce NOW today, adding nine titles total. Ultimate members can stream the 100-player tactical shooter at RTX 5080-class performance without local installs.