← Grok 4.5 · API · High effort
Evidenceacceptedvendor_reported
SWE-bench Pro
SpaceXAI Grok 4.5 launch — SWE-Bench Pro
ev_2026_07_grok45_swe_pro_vendor
Result
64.7%
Confidence B
- Tested
- 2026-07-08
- Runs
- 1
Configuration
How it was measured
- Exact model
- grok-4.5
- Reasoning
- high
- Agent harness
- vendor_swe_pro
- Tool access
- agentic coding tools
- Prompting
- vendor published eval
Secondary STR (vendor).
Caveats
Known limitations
- Vendor-reported; not mini-swe-agent Verified.
Weighting
Evidence weight factors
The engine weights each evidence row by these factors (0–1) when fusing scores — higher is more trusted.
relevance0.85
quality0.55
recency1.00
health0.80
comparability0.65
reliability0.70
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 →