# AdvancedIF — Mercor evaluation (AdvancedIF (Mercor))

> Following complex instructions graded against rubric criteria. This table shows Mercor-run configurations for reference and is excluded from model rankings.

Canonical page: https://benchlm.ai/benchmarks/mercoradvancedif

- Category: [Instruction Following](/instruction-following)
- Last updated: October 15, 2026 capture

Evaluation results published by [Mercor](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/). Published evaluation aggregates only; task contents and dataset license grants are not included.

We captured Mercor's AdvancedIF results on October 15, 2026. The table preserves 45 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.

The headline is the source page's selected Pass@1 metric. The task set, harness, tools, and grader belong to this Mercor run and may differ from another evaluator's results.

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 llm-as-a-judge (lmaaj) over expert-written rubric criteria and a judge of Gemini 3.5 Flash. Its run-count label is k = 4. The linked result page takes precedence where its setup differs, as noted above.

- [Mercor AdvancedIF results](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/)
- [Mercor APEX benchmark index](https://www.mercor.com/apex/)
- [Mercor methodology](https://www.mercor.com/apex/methodology/)
- [Source paper](https://arxiv.org/abs/2511.10507)
- [Upstream dataset](https://huggingface.co/datasets/facebook/AdvancedIF)

## About AdvancedIF (Mercor)

- Tasks: 1,645 source-reported tasks
- Format: Pass@1
- Difficulty: Source-specific evaluation
- Paper: [AdvancedIF evaluation results](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-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.

AdvancedIF (Mercor) is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (45 configurations)

| Rank | Model | Configuration | Creator | Score |
|------|-------|---------------|---------|-------|
| 1 | [Gemini 3.1 Pro](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · high reasoning · Published margin ±1.4 · 6,580 reported samples | Google | 86.70% |
| 2 | [Gemini 3.6 Flash](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · high reasoning · Published margin ±1.4 · 6,580 reported samples | Google | 85.30% |
| 3 | [Qwen3.8-Max](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · xhigh reasoning · Published margin ±1.4 · 6,571 reported samples | Alibaba | 85.10% |
| 4 | [Fable 5.1](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.5 · 6,494 reported samples | Anthropic | 84.20% |
| 5 | [DeepSeek-V4.1-Flash](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.4 · 6,580 reported samples | DeepSeek | 84.00% |
| 6 | [Sonnet 5.5](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.5 · 6,553 reported samples | Anthropic | 83.40% |
| 7 | [Opus 5.5](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.6 · 6,580 reported samples | Anthropic | 82.80% |
| 8 | [Fable 5.1](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · high reasoning · Published margin ±1.5 · 6,580 reported samples | Anthropic | 82.60% |
| 9 | [GPT-6.1 Sol](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.6 · 6,557 reported samples | OpenAI | 82.40% |
| 10 | [GPT-5.6 Sol (Pro)](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.6 · 6,580 reported samples | OpenAI | 82.10% |
| 11 | [GPT-6 Astra](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.6 · 6,580 reported samples | OpenAI | 82.00% |
| 12 | [GPT-5.5](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · xhigh reasoning · Published margin ±1.6 · 6,580 reported samples | OpenAI | 81.30% |
| 13 | [GPT-6 Astra](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · xhigh reasoning · Published margin ±1.7 · 6,571 reported samples | OpenAI | 81.30% |
| 14 | [GPT-5.4](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · xhigh reasoning · Published margin ±1.6 · 6,580 reported samples | OpenAI | 81.00% |
| 15 | [GPT-6 Sol](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.6 · 6,569 reported samples | OpenAI | 80.90% |
| 16 | [Grok 4.6](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · xhigh reasoning · Published margin ±1.6 · 6,578 reported samples | xAI | 80.90% |
| 17 | [Haiku 5.5](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.6 · 6,580 reported samples | Anthropic | 80.80% |
| 18 | [Opus 5](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.6 · 6,580 reported samples | Anthropic | 80.50% |
| 19 | [Grok 4.7](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · xhigh reasoning · Published margin ±1.5 · 6,578 reported samples | xAI | 80.20% |
| 20 | [Muse Spark 1.1](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · xhigh reasoning · Published margin ±1.6 · 6,570 reported samples | Meta | 79.40% |
| 21 | [Grok 4.5](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · high reasoning · Published margin ±1.6 · 6,580 reported samples | xAI | 79.30% |
| 22 | [GLM-5.3](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.7 · 6,558 reported samples | Zhipu | 78.20% |
| 23 | [GPT-5.6 Terra](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.7 · 6,580 reported samples | OpenAI | 78.10% |
| 24 | [Qwen3.5](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · Published margin ±1.6 · 6,580 reported samples | Alibaba | 78.10% |
| 25 | [Inkling](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · high reasoning · Published margin ±1.6 · 6,580 reported samples | Thinking Machines | 77.60% |
| 26 | [GLM-5.2](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.6 · 6,580 reported samples | Zhipu | 77.40% |
| 27 | [Gemma 4 31B](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · Published margin ±1.7 · 6,580 reported samples | Google | 77.00% |
| 28 | [DeepSeek-V4-Flash](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.7 · 6,580 reported samples | DeepSeek | 76.60% |
| 29 | [GPT-5.6 Luna](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.7 · 6,580 reported samples | OpenAI | 76.00% |
| 30 | [GPT-6 Luna](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.8 · 6,580 reported samples | OpenAI | 74.90% |
| 31 | [DeepSeek-V4-Pro](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.7 · 6,535 reported samples | DeepSeek | 74.70% |
| 32 | [Kimi K2.7 Code](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · high reasoning · Published margin ±1.7 · 6,580 reported samples | Kimi | 73.80% |
| 33 | [Opus 4.8](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.8 · 6,580 reported samples | Anthropic | 73.70% |
| 34 | [Gemini 3.5 Flash](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · high reasoning · Published margin ±1.7 · 6,580 reported samples | Google | 73.70% |
| 35 | [Sonnet 5](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.8 · 6,557 reported samples | Anthropic | 73.30% |
| 36 | [Opus 4.7](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.9 · 6,580 reported samples | Anthropic | 73.10% |
| 37 | [Sonnet 4.6](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · high reasoning · Published margin ±1.9 · 6,580 reported samples | Anthropic | 70.80% |
| 38 | [Kimi K2 (Thinking)](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · high reasoning · Published margin ±1.8 · 6,580 reported samples | Kimi | 68.50% |
| 39 | [GPT-OSS-120B](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · high reasoning · Published margin ±1.8 · 6,580 reported samples | OpenAI | 68.30% |
| 40 | [Fable 5](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.9 · 6,580 reported samples | Anthropic | 65.20% |
| 41 | [Kimi K3](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · max reasoning · Published margin ±1.8 · 6,557 reported samples | Kimi | 58.70% |
| 42 | [MiniMax-M3](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · high reasoning · Published margin ±1.9 · 6,580 reported samples | MiniMax | 56.30% |
| 43 | [Nemotron 3 Ultra](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · high reasoning · Published margin ±1.9 · 6,549 reported samples | NVIDIA | 53.70% |
| 44 | [DeepSeek-V3.2](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · Published margin ±2 · 6,576 reported samples | DeepSeek | 51.30% |
| 45 | [MiniMax-M2.7](https://www.mercor.com/apex/oss-benchmarks/oss-advanced-if-leaderboard/) | Single-shot agent · high reasoning · Published margin ±2 · 6,580 reported samples | MiniMax | 47.80% |

## FAQ

### What does AdvancedIF (Mercor) measure?

Following complex instructions graded against rubric criteria. This table shows Mercor-run configurations for reference and is excluded from model rankings.

### Which model leads the published AdvancedIF (Mercor) snapshot?

Gemini 3.1 Pro currently leads the published AdvancedIF (Mercor) snapshot with a score of 86.70%.

### How many models are evaluated on AdvancedIF (Mercor)?

The October 15, 2026 capture contains 45 source configurations.

### Does AdvancedIF (Mercor) affect BenchLM's overall score?

Not directly. AdvancedIF (Mercor) is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.