# CharXiv — Mercor evaluation (CharXiv (Mercor))

> Answering descriptive and reasoning questions about research-paper charts. This table shows Mercor-run configurations for reference and is excluded from model rankings.

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

- Category: [Multimodal & Grounded](/multimodal-grounded)
- Last updated: October 15, 2026 capture

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

We captured Mercor's CharXiv results on October 15, 2026. The table preserves 40 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) and a judge of Gemini 3.5 Flash. Its run-count label is k = 3. The linked result page takes precedence where its setup differs, as noted above.

- [Mercor CharXiv results](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/)
- [Mercor APEX benchmark index](https://www.mercor.com/apex/)
- [Mercor methodology](https://www.mercor.com/apex/methodology/)
- [Source paper](https://arxiv.org/abs/2406.18521)
- [Upstream evaluation code](https://github.com/princeton-nlp/CharXiv)
- [Upstream dataset](https://huggingface.co/datasets/princeton-nlp/CharXiv)

## About CharXiv (Mercor)

- Tasks: 5,000 source-reported tasks
- Format: Pass@1
- Difficulty: Source-specific evaluation
- Paper: [CharXiv evaluation results](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-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.

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

## Leaderboard (40 configurations)

| Rank | Model | Configuration | Creator | Score |
|------|-------|---------------|---------|-------|
| 1 | [Opus 5.5](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.5 · 15,000 reported samples | Anthropic | 96.40% |
| 2 | [GPT-6 Astra](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.5 · 15,000 reported samples | OpenAI | 96.10% |
| 3 | [Sonnet 5.5](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.5 · 14,999 reported samples | Anthropic | 96.10% |
| 4 | [GPT-6.1 Sol](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.5 · 14,986 reported samples | OpenAI | 96.00% |
| 5 | [GPT-6 Astra](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · xhigh reasoning · Published margin ±0.5 · 15,000 reported samples | OpenAI | 95.80% |
| 6 | [Gemini 3.8 Flash](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · high reasoning · Published margin ±0.5 · 15,000 reported samples | Google | 95.40% |
| 7 | [Gemini 3.7 Flash](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · high reasoning · Published margin ±0.5 · 15,000 reported samples | Google | 95.30% |
| 8 | [Muse Spark 1.2](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · xhigh reasoning · Published margin ±0.5 · 15,000 reported samples | Meta | 95.30% |
| 9 | [Muse Spark 1.1](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · xhigh reasoning · Published margin ±0.5 · 15,000 reported samples | Meta | 95.20% |
| 10 | [Gemini 3.6 Flash](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · high reasoning · Published margin ±0.6 · 15,000 reported samples | Google | 95.10% |
| 11 | [Fable 5.1](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.6 · 15,000 reported samples | Anthropic | 95.10% |
| 12 | [GPT-6 Sol](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.6 · 15,000 reported samples | OpenAI | 95.00% |
| 13 | [Opus 5](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.6 · 14,998 reported samples | Anthropic | 94.60% |
| 14 | [GPT-5.5](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · xhigh reasoning · Published margin ±0.6 · 14,998 reported samples | OpenAI | 94.60% |
| 15 | [Gemini 3.1 Pro](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · high reasoning · Published margin ±0.6 · 14,998 reported samples | Google | 94.60% |
| 16 | [Gemini 3.5 Flash](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · high reasoning · Published margin ±0.6 · 14,999 reported samples | Google | 94.30% |
| 17 | [Qwen3.8-Max](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · xhigh reasoning · Published margin ±0.6 · 15,000 reported samples | Alibaba | 94.10% |
| 18 | [GPT-5.6 Terra](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.6 · 14,998 reported samples | OpenAI | 94.10% |
| 19 | [Fable 5.1](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · high reasoning · Published margin ±0.6 · 15,000 reported samples | Anthropic | 94.10% |
| 20 | [Haiku 5.5](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.6 · 15,000 reported samples | Anthropic | 94.10% |
| 21 | [Kimi K3](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.6 · 15,000 reported samples | Kimi | 93.90% |
| 22 | [Muse Spark 1.3](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.6 · 14,993 reported samples | Meta | 93.90% |
| 23 | [GPT-5.4](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · xhigh reasoning · Published margin ±0.6 · 14,998 reported samples | OpenAI | 93.40% |
| 24 | [Opus 4.8](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.6 · 14,991 reported samples | Anthropic | 93.20% |
| 25 | [GPT-5.6 Luna](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.6 · 15,000 reported samples | OpenAI | 92.90% |
| 26 | [Grok 4.6](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · xhigh reasoning · Published margin ±0.6 · 15,000 reported samples | xAI | 92.90% |
| 27 | [Grok 4.5](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · high reasoning · Published margin ±0.7 · 14,999 reported samples | xAI | 92.80% |
| 28 | [Opus 4.7](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.7 · 14,999 reported samples | Anthropic | 92.60% |
| 29 | [GLM-5.3-Flash](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.7 · 14,998 reported samples | Zhipu | 92.50% |
| 30 | [GPT-5.6 Sol (Pro)](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.6 · 14,998 reported samples | OpenAI | 92.10% |
| 31 | [Grok 4.7](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · xhigh reasoning · Published margin ±0.7 · 14,648 reported samples | xAI | 92.00% |
| 32 | [Qwen3.5](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · Published margin ±0.7 · 14,995 reported samples | Alibaba | 91.60% |
| 33 | [GPT-6 Luna](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.7 · 15,000 reported samples | OpenAI | 90.90% |
| 34 | [Sonnet 4.6](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · high reasoning · Published margin ±0.8 · 14,999 reported samples | Anthropic | 90.50% |
| 35 | [DeepSeek-V4.1-Flash](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.7 · 15,000 reported samples | DeepSeek | 90.10% |
| 36 | [Fable 5](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.8 · 20,000 reported samples | Anthropic | 89.10% |
| 37 | [Inkling](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · high reasoning · Published margin ±0.8 · 15,000 reported samples | Thinking Machines | 88.90% |
| 38 | [Sonnet 5](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±0.7 · 14,997 reported samples | Anthropic | 88.80% |
| 39 | [MiniMax-M3](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · high reasoning · Published margin ±0.8 · 15,000 reported samples | MiniMax | 85.90% |
| 40 | [GLM-5.3](https://www.mercor.com/apex/oss-benchmarks/oss-charxiv-leaderboard/) | Single-shot multimodal · max reasoning · Published margin ±1.1 · 14,965 reported samples | Zhipu | 63.00% |

## FAQ

### What does CharXiv (Mercor) measure?

Answering descriptive and reasoning questions about research-paper charts. This table shows Mercor-run configurations for reference and is excluded from model rankings.

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

Opus 5.5 currently leads the published CharXiv (Mercor) snapshot with a score of 96.40%.

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

The October 15, 2026 capture contains 40 source configurations.

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

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