← Claude Opus 4.8 · API · High effort (default)
Evidenceacceptedvendor_reported
SWE-bench Pro
Claude Opus 4.8 System Card — SWE-bench Pro
ev_2026_07_opus48_swe_pro
Result
69.2%
Confidence B
- Tested
- 2026-05-28
- Runs
- 1
Configuration
How it was measured
- Exact model
- claude-opus-4-8
- Reasoning
- high
- Agent harness
- anthropic_internal_swe_pro_eval
- Tool access
- agentic coding tools (vendor)
- Prompting
- vendor system-card protocol
SECONDARY STR — vendor SWE-bench Pro.
Caveats
Known limitations
- Vendor-reported; not mini-SWE-agent standardised.
Weighting
Evidence weight factors
The engine weights each evidence row by these factors (0–1) when fusing scores — higher is more trusted.
relevance0.95
quality0.55
recency0.95
health0.85
comparability0.75
reliability0.65
Benchmark
SWE-bench Pro
Harder SWE-bench variant with multi-file diffs from actively maintained repositories; used by vendors and labs as an agentic coding stress test.
Benchmark source →