MQ
MODELQUEST
choose the right AI for the quest
DeepSeek V4 Pro · API · Max effort
Evidenceacceptedthird_party_leaderboard

SWE-bench Verified

Vals.ai SWE-bench Verified (mini-swe-agent) — DeepSeek V4

ev_2026_07_dsv4pro_swe_vals

Result
77.4%
Confidence A
Tested
2026-07-09
Runs
1
Configuration

How it was measured

Exact model
deepseek-v4-pro
Reasoning
max
Agent harness
mini-swe-agent
Tool access
bash only (no specialized editor tools)
Prompting
Vals default provider config; max tokens raised to model max

PRIMARY STR — independent Vals.ai SWE-bench Verified board using mini-swe-agent for all models. Prefer over openlm mirror and vendor for comparative STR.

Caveats

Known limitations

  • Point estimate with published stderr ±1.87; multi-trial CI not fully transcribed.
  • Vals labels "DeepSeek V4" (Pro line · max effort) — not Flash.
  • mini-swe-agent is intentionally minimal; richer harnesses can score differently.
  • Do not equate with vendor SWE Verified 80.6% (different harness/config).
Weighting

Evidence weight factors

The engine weights each evidence row by these factors (0–1) when fusing scores — higher is more trusted.

relevance0.95
quality0.85
recency1.00
health0.75
comparability0.90
reliability0.85
Benchmark

SWE-bench Verified

Human-filtered 500-instance subset of SWE-bench evaluating whether models can resolve real GitHub issues in existing repositories under a controlled agent harness.

Benchmark source →