# APEX-SWE (Inspect) — Mercor evaluation (APEX-SWE (Inspect) (Mercor)) > Software integration and production-failure diagnosis with the Inspect harness. This table shows Mercor-run configurations for reference and is excluded from model rankings. Canonical page: https://benchlm.ai/benchmarks/mercorapexsweinspect - Category: [Coding](/coding) - Last updated: October 15, 2026 capture Evaluation results published by [Mercor](https://www.mercor.com/apex/apex-swe-leaderboard/). Published evaluation aggregates only; task contents and dataset license grants are not included. We captured Mercor's APEX-SWE (Inspect) results on October 15, 2026. The table preserves 23 published configurations, their source identifiers, reasoning settings, and disclosed pass@1 values. These are results from Mercor's evaluation setup; we did not rerun them. This dated source table is display only and does not enter overall or category rankings. Source configurations retain their external identity, including Pro descriptors and different effort settings. These are the disclosed Inspect Pass@1 results across integration and observability work. Terminus-2 is a separate harness profile. One source record labels its metric pass@1 rather than the declared pass-1 selector; it is retained in private evidence and withheld from this table until the source resolves that mismatch. The row labels retain published error margins and sample counts when supplied. Missing margins and counts remain unreported; a sample count is not used to reconstruct a task denominator. Except where the dedicated source defines an interval, the confidence level is unreported. Equal displayed scores and overlapping margins do not establish statistical ties. Capture time does not establish individual evaluation dates. The general methodology lists expert-authored rubrics plus repository test suites and a judge of LLM judge + programmatic tests. Its run-count label is k = 3. The linked result page takes precedence where its setup differs, as noted above. - [Mercor APEX-SWE (Inspect) results](https://www.mercor.com/apex/apex-swe-leaderboard/) - [Mercor APEX benchmark index](https://www.mercor.com/apex/) - [Mercor methodology](https://www.mercor.com/apex/methodology/) - [Source paper](https://arxiv.org/pdf/2601.08806) - [Upstream evaluation code](https://github.com/Mercor-Intelligence/apex-swe) - [Upstream dataset](https://huggingface.co/datasets/mercor/APEX-SWE) - [Mercor evaluation notes](https://www.mercor.com/blog/introducing-apex-swe/) ## About APEX-SWE (Inspect) (Mercor) - Tasks: Integration and observability engineering tasks - Format: Pass@1 - Difficulty: Source-specific evaluation - Paper: [APEX-SWE (Inspect) evaluation results](https://www.mercor.com/apex/apex-swe-leaderboard/) The selected headline is Pass@1. Source model identifiers, effort settings, published margins, and reported sample counts remain attached to the source configuration. Public-set results and Mercor task extensions occupy separate tables. APEX-SWE (Inspect) (Mercor) is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula. ## Leaderboard (23 configurations) | Rank | Model | Configuration | Creator | Score | |------|-------|---------------|---------|-------| | 1 | [Fable 5](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · max reasoning · Published margin ±6 · 799 reported samples | Anthropic | 54.83% | | 2 | [Opus 5](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · max reasoning · Published margin ±5.5 · 800 reported samples | Anthropic | 54.67% | | 3 | [Grok 4.5](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · high reasoning · Published margin ±6 · 800 reported samples | xAI | 51.17% | | 4 | [Opus 4.8](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · high reasoning · Published margin ±6 · Sample count not reported | Anthropic | 47.30% | | 5 | [Kimi K3](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · max reasoning · Published margin ±6 · 800 reported samples | Kimi | 44.67% | | 6 | [Sonnet 5](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · high reasoning · Published margin ±6 · Sample count not reported | Anthropic | 43.65% | | 7 | [GPT-5.3-Codex](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · high reasoning · Published margin ±6.25 · Sample count not reported | OpenAI | 41.50% | | 8 | [GPT-5.5](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · xhigh reasoning · Published margin ±6.5 · 1,600 reported samples | OpenAI | 40.84% | | 9 | [GPT-5.6 Sol](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · xhigh reasoning · Published margin ±5.5 · 800 reported samples | OpenAI | 39.67% | | 10 | [Gemini 3.6 Flash](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · high reasoning · Published margin ±6 · 800 reported samples | Google | 39.67% | | 11 | [GLM-5.2](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · Published margin ±5.5 · Sample count not reported | Zhipu | 37.32% | | 12 | [Gemini 3.5 Flash](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · high reasoning · Published margin ±5.8 · 800 reported samples | Google | 32.50% | | 13 | [Kimi K2.7](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · Published margin ±6 · Sample count not reported | Kimi | 32.48% | | 14 | [Cognition SWE-1.6 Preview](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · Published margin ±5.75 · Sample count not reported | Cognition | 31.70% | | 15 | [Sonnet 4.5](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · high reasoning · Published margin ±6 · Sample count not reported | Anthropic | 31.00% | | 16 | [GPT-5.2-Codex](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · high reasoning · Published margin ±5.84 · Sample count not reported | OpenAI | 30.80% | | 17 | [Kimi K2.5](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · Published margin ±5.83 · Sample count not reported | Kimi | 30.30% | | 18 | [GPT-5.1-Codex](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · high reasoning · Published margin ±5.59 · Sample count not reported | OpenAI | 29.00% | | 19 | [Gemini 3.1 Pro](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · high reasoning · Published margin ±5.42 · 800 reported samples | Google | 28.80% | | 20 | [Gemini 3 Pro](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · high reasoning · Published margin ±5.5 · Sample count not reported | Google | 26.80% | | 21 | [Grok 4](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · Published margin ±5 · Sample count not reported | xAI | 21.00% | | 22 | [DeepSeek-V3.2](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · Published margin ±4.33 · 800 reported samples | DeepSeek | 15.80% | | 23 | [Kimi K2 (Thinking)](https://www.mercor.com/apex/apex-swe-leaderboard/) | Inspect · Published margin ±3.58 · Sample count not reported | Kimi | 11.20% | ## FAQ ### What does APEX-SWE (Inspect) (Mercor) measure? Software integration and production-failure diagnosis with the Inspect harness. This table shows Mercor-run configurations for reference and is excluded from model rankings. ### Which model leads the published APEX-SWE (Inspect) (Mercor) snapshot? Fable 5 currently leads the published APEX-SWE (Inspect) (Mercor) snapshot with a score of 54.83%. ### How many models are evaluated on APEX-SWE (Inspect) (Mercor)? The October 15, 2026 capture contains 23 source configurations. ### Does APEX-SWE (Inspect) (Mercor) affect BenchLM's overall score? Not directly. APEX-SWE (Inspect) (Mercor) is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.