# DeepResearch Bench II — Mercor evaluation (DeepResearch Bench II (Mercor))

> Writing research reports from web evidence. This table shows Mercor-run configurations for reference and is excluded from model rankings.

Canonical page: https://benchlm.ai/benchmarks/mercordeepresearchbench2

- Category: [Agentic](/agentic)
- Last updated: October 15, 2026 capture

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

We captured Mercor's DeepResearch Bench II results on October 15, 2026. The table preserves 28 published configurations, their source identifiers, reasoning settings, and disclosed mean score 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 published mean score, which can include partial credit. It must not be read as a strict task-completion rate.

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, scored criterion by criterion and a judge of GPT 5.5. Its run-count label is k = 1. The linked result page takes precedence where its setup differs, as noted above.

- [Mercor DeepResearch Bench II results](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/)
- [Mercor APEX benchmark index](https://www.mercor.com/apex/)
- [Mercor methodology](https://www.mercor.com/apex/methodology/)
- [Source paper](https://arxiv.org/abs/2601.08536)
- [Upstream evaluation code](https://github.com/imlrz/DeepResearch-Bench-II)
- [Upstream dataset](https://huggingface.co/datasets/muset-ai/DeepResearch-Bench-II-Dataset)

## About DeepResearch Bench II (Mercor)

- Tasks: 130 source-reported tasks
- Format: Mean Score
- Difficulty: Source-specific evaluation
- Paper: [DeepResearch Bench II evaluation results](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/)

The selected headline is Mean Score. 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.

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

## Leaderboard (28 configurations)

| Rank | Model | Configuration | Creator | Score |
|------|-------|---------------|---------|-------|
| 1 | [Opus 5](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · high reasoning · Published margin ±3.4 · 130 reported samples | Anthropic | 56.10% |
| 2 | [Fable 5.1](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · max reasoning · Published margin ±3.8 · 124 reported samples | Anthropic | 55.00% |
| 3 | [Sonnet 4.6](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · high reasoning · Published margin ±3.5 · 130 reported samples | Anthropic | 53.30% |
| 4 | [Opus 5.5](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · max reasoning · Published margin ±3.5 · 125 reported samples | Anthropic | 53.30% |
| 5 | [DeepSeek-V4.1-Flash](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · max reasoning · Published margin ±3.3 · 130 reported samples | DeepSeek | 53.20% |
| 6 | [GLM-5.3](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · max reasoning · Published margin ±3.6 · 128 reported samples | Zhipu | 52.00% |
| 7 | [Sonnet 5.5](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · max reasoning · Published margin ±3.3 · 130 reported samples | Anthropic | 51.70% |
| 8 | [GPT 5.6 Sol](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · medium reasoning · Published margin ±3.1 · 130 reported samples | OpenAI | 51.50% |
| 9 | [Qwen3.8-Max](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · xhigh reasoning · Published margin ±3.4 · 128 reported samples | Alibaba | 51.00% |
| 10 | [GPT-6 Astra](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · xhigh reasoning · Published margin ±3.4 · 130 reported samples | OpenAI | 50.10% |
| 11 | [GPT-5.6 Luna](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · max reasoning · Published margin ±3.4 · 129 reported samples | OpenAI | 49.70% |
| 12 | [Grok 4.7](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · xhigh reasoning · Published margin ±3.5 · 127 reported samples | xAI | 49.00% |
| 13 | [GPT 5.6 Terra](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · medium reasoning · Published margin ±3.2 · 130 reported samples | OpenAI | 47.20% |
| 14 | [Grok 4.5](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · high reasoning · Published margin ±3.1 · 130 reported samples | xAI | 47.10% |
| 15 | [Fable 5.1](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · high reasoning · Published margin ±4 · 130 reported samples | Anthropic | 45.90% |
| 16 | [GPT-5.6 Terra](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · max reasoning · Published margin ±3.9 · 130 reported samples | OpenAI | 45.80% |
| 17 | [Opus 4.7](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · high reasoning · Published margin ±3.3 · 130 reported samples | Anthropic | 45.40% |
| 18 | [GPT 5.5](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · medium reasoning · Published margin ±3.2 · 130 reported samples | OpenAI | 45.40% |
| 19 | [Opus 4.8](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · high reasoning · Published margin ±3 · 130 reported samples | Anthropic | 44.60% |
| 20 | [DeepSeek-V4-Pro](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · high reasoning · Published margin ±3.6 · 130 reported samples | DeepSeek | 44.20% |
| 21 | [GPT 5.6 Luna](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · medium reasoning · Published margin ±3 · 130 reported samples | OpenAI | 43.90% |
| 22 | [Gemini 3.7 Flash](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · medium reasoning · Published margin ±2.9 · 130 reported samples | Google | 43.40% |
| 23 | [GPT-6 Luna](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · max reasoning · Published margin ±3 · 130 reported samples | OpenAI | 42.90% |
| 24 | [GPT-6 Sol](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · max reasoning · Published margin ±3.1 · 130 reported samples | OpenAI | 42.90% |
| 25 | [Sonnet 5](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · high reasoning · Published margin ±3.8 · 130 reported samples | Anthropic | 39.70% |
| 26 | [GPT 5.4](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · none reasoning · Published margin ±3.5 · 130 reported samples | OpenAI | 38.40% |
| 27 | [Gemini 3.6 Flash](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · medium reasoning · Published margin ±3.9 · 130 reported samples | Google | 36.30% |
| 28 | [Haiku 5.5](https://www.mercor.com/apex/oss-benchmarks/oss-deepresearch-bench-ii-leaderboard/) | Web research agent (100 max steps) · max reasoning · Published margin ±3 · 128 reported samples | Anthropic | 34.30% |

## FAQ

### What does DeepResearch Bench II (Mercor) measure?

Writing research reports from web evidence. This table shows Mercor-run configurations for reference and is excluded from model rankings.

### Which model leads the published DeepResearch Bench II (Mercor) snapshot?

Opus 5 currently leads the published DeepResearch Bench II (Mercor) snapshot with a score of 56.10%.

### How many models are evaluated on DeepResearch Bench II (Mercor)?

The October 15, 2026 capture contains 28 source configurations.

### Does DeepResearch Bench II (Mercor) affect BenchLM's overall score?

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