Anthropic
Claude Opus 4.7
Previous Anthropic flagship — still elite, especially on science reasoning.
API onlytextimagefileREL 2026-04-16
Compare this model →Verified 2026-06-12
Overview
Claude Opus 4.7 was Anthropic's flagship until Opus 4.8 landed. It remains one of the strongest models on GPQA Diamond (94.2%) and SWE-bench Verified (82.0%), at the same $5/$25 price as its successor — so most users should simply pick 4.8 unless they have a 4.7-tuned workflow.
Benchmark scores
Methodology →Intelligence57
Coding82
Reasoning94.2
Mathn/a
Sources and verification
- Coding: SWE-bench Verified · checked 2026-06-09
- Intelligence: Artificial Analysis Intelligence Index · checked 2026-06-12
- Reasoning: GPQA Diamond 94.2%
- Catalogue and pricing: OpenRouter model record · checked 2026-06-12
Strengths and weaknesses
Strengths
- 94.2% GPQA Diamond — elite science reasoning
- Proven in production agent stacks
- 1M-token context
Trade-offs
- Same price as the newer Opus 4.8
- Closed weights
Reach for this model when
Teams with workflows already tuned on 4.7; otherwise use Opus 4.8.
Run something like it locally
Closest open-weights alternatives: DeepSeek V4 Pro or GLM-5.1 via vLLM.
