Latest stories
-
Stability AI model access shifts following license revisions
Stability AI has introduced a new Community License for models like Stable Diffusion 3.5, allowing free use for organizations with annual revenues…
-
Figure AI and the BMW Spartanburg production deployment
Figure AI’s Figure 02 robots achieved 99 percent placement accuracy at BMW’s Spartanburg plant, supporting the assembly of 30,000 X3 vehicles. The…
-
How to get started with Runway’s Gen-4 Turbo video generation API
Learn how advertising studios can implement Runway’s Gen-4 Turbo API for rapid video production. This guide covers developer portal setup, credit math…
-
DeepSeek-V4 vs Qwen 3: Which open-source reasoning model leads
This comparison evaluates DeepSeek V4 Flash, Qwen3.6 35B A3B, and GLM-4.6 across math, coding, and speed benchmarks. DeepSeek V4 Flash offers the…
-
The sovereign capital shift in AI semiconductor investments
State capital is becoming the primary structural funder for global AI infrastructure. Major moves include Saudi Arabia’s $100 billion HUMAIN entity and…
-
Gemini 2.5 Pro beats o3 high on coding benchmarks
Gemini 2.5 Pro leads in code generation with a 74.2% score on LiveCodeBench, outperforming o3 high. While Gemini excels in coding tasks…
-
Tesla shifts from Dojo to AI6 training clusters
Elon Musk has disbanded the Dojo team in favor of the AI6 chip architecture, which scales from FSD to Optimus. This strategic…
-
Why Cerebras’s Andromeda supercomputer matters for OpenAI
OpenAI has invested $10 billion in Cerebras to address inference bottlenecks for models like GPT-5. The Andromeda supercomputer utilizes wafer architecture to…
-
The $400M foundation for universal robot brains
Physical Intelligence secured $400 million to develop the π0 model, a generalist robotic brain designed to control diverse hardware. This technology aims…
-
Lightmatter’s Passage photonic interconnect and NVIDIA Blackwell
Lightmatter’s Passage M1000 photonic interposer provides 114 Tbps of bidirectional bandwidth to solve chiplet communication bottlenecks. This technology integrates with NVIDIA’s Blackwell…
-
Power delivery failures in SambaNova SN40L rack deployments
Deploying SambaNova SN40L racks requires precise power and thermal validation to avoid instability. Common mistakes include insufficient PDU margin testing and overlooking…
-
Common fine-tuning mistakes that ruin Llama 3.3 enterprise deployments
Avoid deployment failures by managing data quality, quantization precision, and catastrophic forgetting. While Llama 3.3 70B offers cost advantages, its 0.11 completion…