A French Startup Pantsed Nvidia With Free Software
ZML just gave away a free tool that runs your AI models fast on any chip, not just Nvidia’s. It is a small, well-aimed kick at the CUDA lock-in that keeps the whole industry hostage.
ZML just gave away a free tool that runs your AI models fast on any chip, not just Nvidia’s. It is a small, well-aimed kick at the CUDA lock-in that keeps the whole industry hostage.
Gemini 3.5 Pro is reportedly targeting July 17 after months of delays — headlined by a 2M-token context window and an Ultra-tier “Deep Think” mode. What’s real and what’s TBD.
GPT-5.6 went public in three tiers. Sol is the coding flagship with a subagent “ultra” mode; Terra is priced to punch Claude Sonnet 5 in the mouth. What to actually use.
Anthropic is reportedly in early talks with Samsung to build a custom AI chip on a 2nm process. Everyone’s trying to escape the Nvidia tax — few will pull it off.
Grok 4.3 landed on Bedrock at $1.25/$2.50. Here is the honest comparison against Sonnet 5 and GPT-5.5 — and when the cheapest model is the correct one.
OpenAI is putting GPT-5.6 Sol on Cerebras wafer-scale chips at up to 750 tokens a second. When latency stops being the bottleneck, the product changes.
Meituan open-sourced LongCat-2.0 on June 30: 1.6 trillion parameters, trained and served entirely on domestic Chinese silicon. The chip chokehold just got tested.
Anthropic dropped Claude Sonnet 5 on June 30 — near-Opus performance at Sonnet money. Here is what changed, what breaks, and why the timing is not an accident.
OpenAI’s leaked 2025 financials show $13B revenue and a $38.5B loss. The real cash burn was $8B. What the numbers actually mean for AI tool prices.
MiniMax M3 undercut GPT-5.5 and Claude Opus on price. A skeptical read on what is real, what is launch-week theatre, and whether you should switch.