China just open-sourced a compact OCR model that can process a full 100 page PDF in a single pass.
It’s called Unlimited OCR. Only 3B parameters. Runs entirely on your own hardware. Most OCR tools still slice documents page by page and lose context across tables, formulas, and
DeepSeek just gave its cheapest model EYES.
V4-Flash-Vision-Exp: Vision at V4-Flash pricing (nothing changed) ✅ Images billed as tokens — max 384 tokens per pic ✅ 1M context, output from $0.66/1M tokens ✅ Text & coding abilities: zero regression
Deepseek just gave its fast agent model eyes.
And the benchmark jump is bigger than I expected. What changed: → DeepSeek V4 Flash Vision X can natively inspect screenshots, charts, designs, documents, and UIs → Images go directly into its reasoning process instead of being
Official news! OpenAI announced the results of its first special chip for running artificial…
Official news! OpenAI announced the results of its first special chip for running artificial intelligence models (Inference), which bears the name “Jalapeño” 🌶️. The most prominent numbers and details they revealed: • The chip offers higher throughput and faster response time at the same time, without having to sacrifice one of them as happens in systems…
You have reached the end of the archive
All of deepseek