# HLE Extended — Mercor evaluation (HLE Extended (Mercor))

> Answering difficult academic questions across many subjects. This table shows Mercor-run configurations for reference and is excluded from model rankings.

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

- Category: [Knowledge](/knowledge)
- Last updated: October 15, 2026 capture

Evaluation results published by [Mercor](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/). Published evaluation aggregates only; task contents and dataset license grants are not included.

We captured Mercor's HLE Extended results on October 15, 2026. The table preserves 31 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.

The headline uses the 447 Mercor-built tasks. The paired original-set values refer to the separately listed 2158-task public benchmark. Missing baselines remain unreported. We keep the extension and original set separate from their upstream benchmark-owner and provider-run tables.

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) and a judge of Gemini 3.5 Flash. Its run-count label is k = 3. Those notes describe the original public benchmark; extension-specific changes are not established by that reference.

- [Mercor HLE Extended results](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/)
- [Mercor APEX benchmark index](https://www.mercor.com/apex/)
- [Mercor methodology](https://www.mercor.com/apex/methodology/)
- [Paired Mercor original-set results](https://www.mercor.com/apex/oss-benchmarks/oss-hle-leaderboard/)
- [Source paper](https://arxiv.org/abs/2501.14249)
- [Upstream evaluation code](https://github.com/centerforaisafety/hle)
- [Upstream dataset](https://huggingface.co/datasets/cais/hle)

## About HLE Extended (Mercor)

- Tasks: 447 Mercor tasks; 2158 tasks in the separately reported original set
- Format: Pass@1
- Difficulty: Source-specific evaluation
- Paper: [HLE Extended evaluation results](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-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.

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

## Leaderboard (31 configurations)

| Rank | Model | Configuration | Creator | Score |
|------|-------|---------------|---------|-------|
| 1 | [GPT-5.6 Sol (Pro)](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · max reasoning · Published margin ±3.7 · Original set not reported · 1,325 reported samples | OpenAI | 77.30% |
| 2 | [GPT-5.6 Terra](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · max reasoning · Published margin ±3.9 · Original set 44.9% ±2 · 1,341 reported samples | OpenAI | 70.70% |
| 3 | [Opus 5](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · high reasoning · Published margin ±4 · Original set 52% ±2 · 1,341 reported samples | Anthropic | 61.00% |
| 4 | [Fable 5](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · high reasoning · Published margin ±4.1 · Original set 51.5% ±2 · 1,341 reported samples | Anthropic | 58.70% |
| 5 | [GPT-5.6 Luna](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · max reasoning · Published margin ±4 · Original set 39.6% ±1.9 · 1,319 reported samples | OpenAI | 57.50% |
| 6 | [GPT-5.4](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · xhigh reasoning · Published margin ±3.8 · Original set 44.1% ±2 · 1,341 reported samples | OpenAI | 56.10% |
| 7 | [GPT-6 Luna](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · max reasoning · Published margin ±4 · Original set 38.7% ±1.9 · 1,341 reported samples | OpenAI | 53.30% |
| 8 | [Opus 4.8](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · high reasoning · Published margin ±4 · Original set 44.8% ±1.9 · 1,341 reported samples | Anthropic | 49.40% |
| 9 | [Qwen3.8-Max](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · xhigh reasoning · Published margin ±3.8 · Original set 42.2% ±1.9 · 1,341 reported samples | Alibaba | 47.10% |
| 10 | [Grok 4.6](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · xhigh reasoning · Published margin ±3.8 · Original set 41.3% ±1.9 · 1,341 reported samples | xAI | 45.80% |
| 11 | [DeepSeek-V4.1-Flash](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · max reasoning · Published margin ±3.7 · Original set 40.5% ±1.9 · 1,339 reported samples | DeepSeek | 45.30% |
| 12 | [Gemini 3.8 Flash](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · high reasoning · Published margin ±3.9 · Original set 47% ±2 · 1,341 reported samples | Google | 36.80% |
| 13 | [DeepSeek-V4-Flash](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · max reasoning · Published margin ±3.5 · Original set 39% ±1.8 · 1,340 reported samples | DeepSeek | 36.10% |
| 14 | [Kimi K3](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · max reasoning · Published margin ±3.3 · Original set 46.5% ±2 · 1,341 reported samples | Kimi | 35.00% |
| 15 | [Gemini 3.7 Flash](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · high reasoning · Published margin ±3.5 · Original set 46% ±1.9 · 1,341 reported samples | Google | 31.60% |
| 16 | [Opus 4.7](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · high reasoning · Published margin ±3.5 · Original set 42.2% ±1.9 · 1,341 reported samples | Anthropic | 31.20% |
| 17 | [Sonnet 5](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · high reasoning · Published margin ±3.7 · Original set 36.3% ±1.8 · 1,341 reported samples | Anthropic | 29.10% |
| 18 | [GLM-5.2](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · max reasoning · Published margin ±3.1 · Original set 31.7% ±1.6 · 1,341 reported samples | Zhipu | 25.80% |
| 19 | [Grok 4.5](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · high reasoning · Published margin ±3.1 · Original set 41.9% ±1.9 · 1,341 reported samples | xAI | 25.40% |
| 20 | [Gemini 3.1 Pro](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · high reasoning · Published margin ±2.6 · Original set 44.1% ±1.9 · 1,341 reported samples | Google | 19.60% |
| 21 | [Gemini 3.5 Flash](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · high reasoning · Published margin ±2.5 · Original set 40.6% ±1.9 · 1,341 reported samples | Google | 14.80% |
| 22 | [Kimi K2.7 Code](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · high reasoning · Published margin ±2.4 · Original set 34% ±1.8 · 1,341 reported samples | Kimi | 13.70% |
| 23 | [Gemini 3.6 Flash](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · high reasoning · Published margin ±2.3 · Original set 41.2% ±1.9 · 1,341 reported samples | Google | 13.30% |
| 24 | [Sonnet 4.6](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · high reasoning · Published margin ±2.4 · Original set 24.4% ±1.6 · 1,341 reported samples | Anthropic | 10.90% |
| 25 | [DeepSeek-V3.2](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · Published margin ±1.9 · Original set 22.3% ±1.5 · 1,341 reported samples | DeepSeek | 8.10% |
| 26 | [GPT-OSS-120B](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · high reasoning · Published margin ±1.9 · Original set 18.5% ±1.4 · 1,341 reported samples | OpenAI | 8.10% |
| 27 | [MiniMax-M2.7](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · high reasoning · Published margin ±1.8 · Original set 27.1% ±1.6 · 1,341 reported samples | MiniMax | 6.60% |
| 28 | [MiniMax-M3](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · high reasoning · Published margin ±1.6 · Original set 37.8% ±1.8 · 1,341 reported samples | MiniMax | 5.40% |
| 29 | [Qwen3.5](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · Published margin ±1.4 · Original set 29.2% ±1.7 · 1,341 reported samples | Alibaba | 4.20% |
| 30 | [Gemma 4 31B](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · Published margin ±1.4 · Original set 23.2% ±1.6 · 1,341 reported samples | Google | 3.80% |
| 31 | [GLM-5 (Thinking)](https://www.mercor.com/apex/mercor-extended/ots-held-out-hle-leaderboard/) | Mercor extended-set run · Published margin ±0.9 · Original set not reported · 1,341 reported samples | Zhipu | 2.00% |

## FAQ

### What does HLE Extended (Mercor) measure?

Answering difficult academic questions across many subjects. This table shows Mercor-run configurations for reference and is excluded from model rankings.

### Which model leads the published HLE Extended (Mercor) snapshot?

GPT-5.6 Sol (Pro) currently leads the published HLE Extended (Mercor) snapshot with a score of 77.30%.

### How many models are evaluated on HLE Extended (Mercor)?

The October 15, 2026 capture contains 31 source configurations.

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

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