Phi-3-medium-4k-instruct-abliterated-v3 — benchmark results
failspy's abliterated version of Microsoft's Phi-3-medium-4k-instruct, with the refusal direction orthogonalized out of the weights. Provider: Microsoft. Released 2024-05-22. Access: Open.
Unified ELO 1463 ± 54, rank #946 of 1776 rated models, from 7 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard - MuSR | 18.52 | Score | 91.2 |
| Open LLM Leaderboard - BBH | 46.73 | Score | 86.6 |
| Open LLM Leaderboard - MMLU-Pro | 37.78 | Score | 81.3 |
| Open LLM Leaderboard - IFEval | 63.19 | Score | 75.3 |
| Open LLM Leaderboard - GPQA | 8.95 | Score | 73.2 |
| Open LLM Leaderboard - MATH Level 5 | 15.94 | Score | 63.7 |
| Open Korean LLM Leaderboard | 126.05 | Average Score (%) | 53.4 |
Interactive version: theaggregate.ai/model?slug=phi-3-medium-4k-instruct-abliterated-v3 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.