Anthropic has released Claude Opus 5, a new mid-tier model positioned between Sonnet and Fable, offering near-Fable intelligence at half the price of Opus 4.8. This release features a 1M token context window, faster processing in its fast mode, and significantly reduced safety classifier interventions, making it more suitable for autonomous coding tasks. Benchmarks indicate Opus 5 surpasses previous models on coding-specific evaluations like Frontier-Bench and CursorBench, and shows a substantial lead in novel problem-solving on ARC-AGI. Alongside the model update, Claude Code v2.1.219 includes enhanced security features like strict network allowlisting and improved subagent capabilities. AI
IMPACT Sets new SOTA on coding benchmarks, potentially lowering costs for AI-powered coding agents.
RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
Read on dev.to — Claude Code tag →
- Anthropic
- ARC-AGI 3
- Claude Code
- Claude Code v2.1.219
- Claude Fable 5
- Claude Max
- Claude Opus 4.8
- Claude Opus 5
- Claude Pro
- Claude Sonnet
- CursorBench 3.2
- Frontier-Bench v0.1
- OSWorld 2.0
- Zapier AutomationBench
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
