Anthropic
99 stories · first 22 Sep · latest 15 Nov
Ongoing case: AI code review in practiceOngoing case: AI-written mathematics and how it is checkedOngoing case: Agents acting on other people's systemsOngoing case: Agents that cheat or misleadCase: Anthropic's AI-accelerated researchOngoing case: Anthropic's Claude modelsCase: Anthropic's IPOOngoing case: Claude in office toolsOngoing case: Cyber-capable AI modelsOngoing case: Distillation attacks and the defences against themOngoing case: Frontier model API pricesOngoing case: Million-token context windowsCase: Music publishers v. AnthropicOngoing case: OpenAI's modelsOngoing case: The White House and AI rulesOngoing case: Watermarks on AI-made contentOngoing case: What the AI buildout costs and needs
Compare with another company or topic
Numbers we checked
Each links to its source.
- Claude Mythos 5.2 Benchmark score: dedicated sandbox escape evaluation, credentials or files used outside the sandbox (Anthropic's own test) · 10 Nov 3.1%
- Claude Mythos 5.2 Price per million output tokens · 10 Nov $40
- Claude Mythos 5.2 Price per million input tokens · 10 Nov $8
- Claude Fable 5.2 Price per million output tokens · 10 Nov $40
- Claude Fable 5.2 Price per million input tokens · 10 Nov $8
- Claude Sonnet 5.5 Other figure: cache hit price per million tokens · 6 Nov $0.10
- Claude Haiku 4.5 Price per million output tokens · 6 Nov $5
- Claude Haiku 4.5 Price per million input tokens · 6 Nov $1
- Claude Haiku 5.5 Price per million output tokens: prompts over 100,000 tokens · 6 Nov $2.50
- Claude Haiku 5.5 Price per million input tokens: prompts over 100,000 tokens · 6 Nov $0.50
- Claude Haiku 5.5 Price per million output tokens: prompts up to 100,000 tokens · 6 Nov $0.50
- Claude Haiku 5.5 Price per million input tokens: prompts up to 100,000 tokens · 6 Nov $0.10
- Claude Fable 5.1 Price per million output tokens · 6 Nov $50
- Claude Fable 5.1 Price per million input tokens · 6 Nov $10
- Claude Opus 5.5 Price per million output tokens · 6 Nov $20
- Claude Opus 5.5 Price per million input tokens · 6 Nov $4
- Claude Sonnet 5.5 Price per million output tokens · 6 Nov $10
- Claude Sonnet 5.5 Price per million input tokens · 6 Nov $2
- Claude Sonnet 5.5 Other figure: cache read price per million tokens, from 7 Oct 2026 · 6 Nov $0.10
- Claude 3.7 Sonnet Benchmark score: SHADE-Arena overall sabotage success, Claude 3.6 Sonnet as monitor · 31 Oct 27%
- Claude Opus 5.5 Benchmark score: Terminal-Bench 4.0 (as reported by Moonshot) · measured by others · 24 Oct 66.4%
- Claude Haiku 5.5 Benchmark score: Anthropic political evenhandedness evaluation · 15 Oct 99%
- Claude Sonnet 5.5 Benchmark score: Anthropic political evenhandedness evaluation · 15 Oct 99%
- Claude Opus 5.5 Benchmark score: Anthropic political evenhandedness evaluation · 15 Oct 99%
- OSS Scanner Other figure: share of 97 critical and high-severity findings that met Anthropic's CVD bar in expert review (Anthropic's own validation) · 12 Oct 88%
- Claude Haiku 5.5 Other figure: cache read price per million tokens, prompts over 100,000 tokens · 8 Oct $0.05
- Claude Haiku 5.5 Other figure: cache read price per million tokens, prompts up to 100,000 tokens · 8 Oct $0.01
- Claude Sonnet 5.5 Benchmark score: OSWorld 2.1, offline subset · 8 Oct 83.9%
- Claude Haiku 4.5 Benchmark score: OSWorld 2.1, offline subset · 8 Oct 15.7%
- Claude Haiku 5.5 Benchmark score: OSWorld 2.1, offline subset · 8 Oct 72.4%
- Claude Haiku 5.5 Context window: context window · 8 Oct 1,000,000 tokens
- Claude Haiku 4.5 Price per million output tokens: output per million tokens · 8 Oct $5
- Claude Haiku 5.5 Price per million output tokens: output per million tokens, prompts over 100,000 tokens · 8 Oct $2.50
- Claude Haiku 5.5 Price per million output tokens: output per million tokens, prompts up to 100,000 tokens · 8 Oct $0.50
- Claude Haiku 4.5 Price per million input tokens: input per million tokens · 8 Oct $1
- Claude Haiku 5.5 Price per million input tokens: input per million tokens, prompts over 100,000 tokens · 8 Oct $0.50
- Claude Haiku 5.5 Price per million input tokens: input per million tokens, prompts up to 100,000 tokens · 8 Oct $0.10
- Claude Sonnet 5.5 Other figure: cache read price per million tokens · 8 Oct $0.10
- Claude Opus 5.5 Benchmark score: CheatBench cheating rate with Claude Code (lower is better) · measured by others · 5 Oct 11.2%
- claude-mythos-preview Benchmark score: Binary Exploitation (Anthropic internal, 100 tasks) · 1 Oct 6%
- Claude Sonnet 5.5 Benchmark score: Terminal-Bench 4.0 · 30 Sep 70.6%
- Claude Opus 5 Benchmark score: ProgramDistill cumulative workflows · measured by others · 28 Sep 28.8%
- Claude Opus 5.5 Context window · 27 Sep 1,000,000 tokens
All 99 stories
Stories per week
321 Sep to 27 Sep
728 Sep to 4 Oct
185 Oct to 11 Oct
1612 Oct to 18 Oct
1219 Oct to 25 Oct
1426 Oct to 1 Nov
112 Nov to 8 Nov
189 Nov to today
Counts cover only the stories we published, not all AI news. Small numbers can mislead.
Named together with
Made by Anthropic
- Claude
- Claude Haiku 4.5
- Claude Haiku 5.5
- Claude Mythos
- Claude Opus 5
- Claude Opus 5.5
- Claude Sonnet 5.5
- claude-mythos-preview
- AI for science
- Agent Skills
- CLAUDE.md
- Claude Agent SDK
- Claude Code
- Claude Code Projects
- Claude Cowork
- Claude Data
- Claude Duet
- Claude Enterprise
- Claude Managed Agents
- Claude Marketplace
- Claude Max
- Claude Meetings
- Claude Sheets
- Claude Ultra
- Claude Whiteboards
- Claude for Government
- Compliance API
- Cyber Mission
- Cyber Verification Program
- Enterprise Frontier Safeguards
- OSS Scanner
- Personal Finance
- Project Glasswing