Anthropic releases Claude Opus 5 with a 1M-token context window and selectable effort levels

Anthropic/Claude models

Anthropic released Claude Opus 5, which it says comes close to the frontier intelligence of Claude Fable 5 at half the price. It keeps Opus 4.8's $5/$25 per million input/output tokens, reads up to 1M tokens of context and lets customers set an effort level to trade quality against speed and cost. It became the default model on Claude Max.

• Context window 1M tokens, enough for a whole codebase or a long document set in one request; up to 128K output tokens
• Effort levels from low to max: higher effort spends more thinking tokens for better answers
• Within 0.5% of Fable 5 on CursorBench 3.2 at half the cost; beat Fable 5's best OSWorld 2.0 computer-use result at just over a third of the cost
• Fast mode: about 2.5x faster output at twice the price
• Anthropic calls it its most aligned model so far, but still behind Mythos 5 on cyber-exploitation skill, with stronger cyber guardrails