ResearchExplainer
Intrinsic Distribution Matching sharpens text in two-step image models
A new training method scores 94.25% on Chinese long-text rendering at two network evaluations, with small gains elsewhere and one benchmark it loses.
Factory AI
Every ClaudeGPT Si story about Factory AI, newest first: news, analysis, reviews and guides.
No Factory AI stories yet
New stories appear here as soon as they are published. Until then, here is the latest from the newsroom.
Read the latest AI newsLatest from ClaudeGPT Si
GuidesGuide
How to try Mistral Large 4: API, free demos or vLLM
Call it as mistral-large-4 on Mistral's API, use the demos while the free budget lasts, or serve the open weights with vLLM on eight GPUs.
ReviewsComparison
Claude Opus 5.5 vs Sonnet 5.5: $4 vs $2, and what else differs
Anthropic's specs show the two models match on context, output and cutoff. Opus costs twice as much and is slower, and Sonnet cannot turn thinking off.
AppsExplainer
Is Claude Opus 5.5 free? No, it needs Pro or the API
Anthropic's pricing page shows Opus as unavailable on the Free plan. Pro costs $20 a month, and the API charges $4 per million input tokens.
GuidesGuide
Qwen 3.8 how to run locally: vLLM and GGUF steps
Serve Qwen3.8-27B with vLLM on one or two RTX 5090 cards, or use a GGUF quantization, with the settings Alibaba recommends.
AI ModelsExplainer
Is Qwen 3.8 open source? Yes, 27B weights are Apache 2.0
Alibaba published Qwen3.8-27B on Hugging Face under a license that allows commercial use. Other Qwen 3.8 models need their own check.