Anthropic reports progress on provable inference for large language models
Anthropic says FlashProve can generate proofs for Llama 4 Maverick at about 110 tokens per second on one H100 GPU.
Claude adds public beta for message threads to cut agent latency
Anthropic says message threads is in public beta, reducing repeated context resend work and cutting median TTFT about 35% for 500k to 1M tokens.
Anthropic expands AI for Science program to institutions and foundations
Anthropic said it is widening its AI for Science program to 20 foundations, research institutions and groups working across science.
IDC brings Quanta into Claude and expands internal use
IDC says Quanta is now available in Claude for customers, while IDC has rolled Claude out across engineering, product, marketing and sales.
Anthropic adds legal artifacts and connectors for Claude
Anthropic introduced ready-to-use legal artifacts in Claude, plus new connectors, partner-built tools and access-to-justice projects.
Harvey launches Intake artifact for Claude legal workflows
Harvey said its new Claude for Legal artifact connects business users to legal teams in Harvey, with routing, tracking and audit controls.
Anthropic expands GSA deal with 40% discount for government users
Anthropic said its expanded GSA agreement gives government buyers a 40% discount on Claude products and may save $1 billion.
Anthropic Adds Financial Advisor Connectors and Workflow Skills to Claude
Anthropic launched Claude for Financial Advisors, adding connectors and workflow skills for research, meeting prep, compliance checks and documentation.
Anthropic outlines cost controls for Claude Code on Google Cloud
Anthropic says organizations can control, track and prove Claude Code spend on Google Cloud using budgets, telemetry and ROI metrics.
Anthropic study finds AI agents can steer expert judgment
Anthropic says professional analysts followed a randomly assigned AI agent’s lean 86% of the time in a causal-inference study.