
Brief · AI
Anthropic releases Claude Fable 5.2 and Mythos 5.2
Anthropic says two new models, Claude Fable 5.2 and Claude Mythos 5.2, improve coding, reasoning and vision and generate output faster than the 5.1 versions.
News feed
Topics
16 items · Page 1 of 2

Brief · AI
Anthropic says two new models, Claude Fable 5.2 and Claude Mythos 5.2, improve coding, reasoning and vision and generate output faster than the 5.1 versions.

Brief · AI
TechCrunch reports Anthropic warned that it is very close to RSI capability, meaning AI that helps train itself. The item gives no measurements or release details.

Brief · AI
Meta says Muse Spark Open is its most capable US open-weight model. It can run locally on two Apple M5 Ultra Mac Studios using the MLX framework.

Brief · AI
Meta says Muse Spark is the most capable US open-weight model to date and can run on two Apple M5 Ultra Mac Studios using MLX.

Brief · AI
Google released Gemini 4 Argon, which it says competes with the best frontier models. It is rolling out in the Gemini app and Antigravity to paying users, with 50 percent discounted API rates through November.

Analysis · AI
Haiku 5.5's 1M-token window is cheap to try and expensive to repeat. Here are the tier math, a break-even formula for long context vs RAG, and what remains unconfirmed.

Brief · AI
Meta released two larger AI models, Muse Wonder and Muse Wonder Full. The company says they compete with the industry's best. Meta stock rose more than 3%.

Brief · Tech
Ars Technica reports that Epic Games is shifting to an AI-first model. It cites a delayed Marvel and Star Wars shooter, where the report says hand-written code and AI-free art will no longer be used.

Analysis · AI
OpenAI's GPT-6 can answer with working charts, forms and calculators instead of paragraphs. Here is what the 44% Instant speedup does and doesn't prove.

Brief · AI
Priced at $0.10 per million input tokens and $0.50 per million output for prompts under 100k tokens.

Brief · AI
Answers can arrive as working charts, forms and calculators instead of paragraphs. OpenAI says the Instant variant answers web-search questions 44% sooner than GPT-5.6 Instant.

Analysis · AI
AI labs have scraped the entire internet and are rapidly running out of human text. Here is why training next-gen LLMs on synthetic data triggers an irreversible degradation called Model Collapse.