Anthropic released Claude Opus 4.7 on April 16, two months after Opus 4.6. The model takes measurable leads on agentic coding, scaled tool use, computer use, and financial-analysis benchmarks against GPT-5.4 and Gemini 3.1 Pro. Pricing stays at $5 input and $25 output per million tokens.
for the same input compared to 4.6. The sticker price is unchanged but effective cost per prompt rises for some workloads. Budget-sensitive API operations should re-benchmark before migrating.
GitHub Copilot shipped 4.7 availability on the same day. Anthropic also rolled out “Fast Mode” on Opus, a beta option at 6x rates ($30 input, $150 output) for latency discount remains 50% across all models; prompt caching drops cache-hit input cost to 10% of standard.
Claude Code, Anthropic’s CLI coding agent, picked up 4.7 immediately. The 1M token context window tokenizer output.
Sources
Primary and corroborating references used for this news item.