AIME 2025 — Mercor evaluation (AIME 2025 (Mercor))

We show this table for reference; we do not rank on it.

Solving competition math problems with integer answers. This table shows Mercor-run configurations for reference and is excluded from model rankings.

Evaluation results published by Mercor. Published evaluation aggregates only; task contents and dataset license grants are not included.

Pass@1 on AIME 2025 (Mercor) — October 15, 2026 capture

We mirror the published pass@1 view for AIME 2025 (Mercor). Opus 5 leads the public snapshot at 100.00%, followed by Fable 5 (100.00%) and DeepSeek-V4-Flash (100.00%). We do not use these results to rank models overall.

Opus 5

Anthropic

Single-shot agent · max reasoning

Published margin ±0 · 120 reported samples

100.00%

Fable 5

Anthropic

Single-shot agent · max reasoning

Published margin ±0 · 120 reported samples

100.00%

DeepSeek-V4-Flash

DeepSeek

Single-shot agent · max reasoning

Published margin ±0 · 120 reported samples

100.00%

49 configurationsMathematicsDated Mercor configurationsDisplay onlyUpdated October 15, 2026 capture

Pass@1 table (49 configurations)

Pass@1 results for AIME 2025 (Mercor)
Rank	Model / configuration	Score	Parameters (B)	Open / closed
1	Opus 5AnthropicSingle-shot agent · max reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
2	Fable 5AnthropicSingle-shot agent · max reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
3	DeepSeek-V4-FlashDeepSeekSingle-shot agent · max reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
4	GPT-5.5OpenAISingle-shot agent · xhigh reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
5	GPT-5.6 Sol (Pro)OpenAISingle-shot agent · max reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
6	GPT-5.6 TerraOpenAISingle-shot agent · max reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
7	Muse Spark 1.1MetaSingle-shot agent · xhigh reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
8	GPT-6 AstraOpenAISingle-shot agent · xhigh reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
9	Fable 5.1AnthropicSingle-shot agent · high reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
10	GPT-6 LunaOpenAISingle-shot agent · max reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
11	Grok 4.7xAISingle-shot agent · xhigh reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
12	GPT-6 SolOpenAISingle-shot agent · max reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
13	Opus 5.5AnthropicSingle-shot agent · max reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
14	Sonnet 5.5AnthropicSingle-shot agent · max reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
15	GLM-5.3ZhipuSingle-shot agent · max reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
16	Fable 5.1AnthropicSingle-shot agent · max reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
17	GPT-6.1 SolOpenAISingle-shot agent · max reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
18	Grok 4.6xAISingle-shot agent · xhigh reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
19	Muse Spark 1.3MetaSingle-shot agent · xhigh reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
20	DeepSeek-V4-Pro-0813DeepSeekSingle-shot agent · max reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
21	Haiku 5.5AnthropicSingle-shot agent · max reasoningPublished margin ±0 · 120 reported samples	100.00%	Not reported	Unknown
22	Muse Spark 1.3MetaSingle-shot agent · max reasoningPublished margin ±0 · 116 reported samples	100.00%	Not reported	Unknown
23	GPT-5.6 LunaOpenAISingle-shot agent · max reasoningPublished margin ±1.3 · 120 reported samples	99.20%	Not reported	Unknown
24	GPT-5.4OpenAISingle-shot agent · xhigh reasoningPublished margin ±1.3 · 120 reported samples	99.20%	Not reported	Unknown
25	Qwen3.8-MaxAlibabaSingle-shot agent · xhigh reasoningPublished margin ±1.3 · 119 reported samples	99.20%	Not reported	Unknown
26	Opus 4.8AnthropicSingle-shot agent · max reasoningPublished margin ±2.5 · 120 reported samples	98.30%	Not reported	Unknown
27	Grok 4.5xAISingle-shot agent · high reasoningPublished margin ±2.5 · 120 reported samples	98.30%	Not reported	Unknown
28	DeepSeek-V4.1-FlashDeepSeekSingle-shot agent · max reasoningPublished margin ±2.1 · 120 reported samples	98.30%	Not reported	Unknown
29	DeepSeek-V4-ProDeepSeekSingle-shot agent · max reasoningPublished margin ±2.5 · 111 reported samples	98.10%	Not reported	Unknown
30	Kimi K2.7 CodeKimiSingle-shot agent · high reasoningPublished margin ±3.3 · 109 reported samples	96.90%	Not reported	Unknown
31	Gemini 3.1 ProGoogleSingle-shot agent · high reasoningPublished margin ±5 · 120 reported samples	96.70%	Not reported	Unknown
32	Kimi K2 (Thinking)KimiSingle-shot agent · high reasoningPublished margin ±5 · 104 reported samples	96.70%	Not reported	Unknown
33	GLM-5.2ZhipuSingle-shot agent · max reasoningPublished margin ±3.7 · 112 reported samples	96.70%	Not reported	Unknown
34	Gemini 3.8 FlashGoogleSingle-shot agent · high reasoningPublished margin ±5 · 120 reported samples	96.70%	Not reported	Unknown
35	Gemini 3.7 FlashGoogleSingle-shot agent · high reasoningPublished margin ±5 · 120 reported samples	96.70%	Not reported	Unknown
36	Gemini 3.6 FlashGoogleSingle-shot agent · high reasoningPublished margin ±5.8 · 120 reported samples	95.80%	Not reported	Unknown
37	Kimi K3KimiSingle-shot agent · max reasoningPublished margin ±3.8 · 120 reported samples	95.80%	Not reported	Unknown
38	GPT-OSS-120BOpenAISingle-shot agent · high reasoningPublished margin ±5.4 · 119 reported samples	94.20%	Not reported	Unknown
39	Opus 4.7AnthropicSingle-shot agent · max reasoningPublished margin ±8.3 · 120 reported samples	93.30%	Not reported	Unknown
40	Nemotron 3 UltraNVIDIASingle-shot agent · high reasoningPublished margin ±8.3 · 98 reported samples	93.30%	Not reported	Unknown
41	InklingThinking MachinesSingle-shot agent · high reasoningPublished margin ±7.1 · 120 reported samples	93.30%	Not reported	Unknown
42	DeepSeek-V3.2DeepSeekSingle-shot agentPublished margin ±7.9 · 120 reported samples	93.30%	Not reported	Unknown
43	MiniMax-M3MiniMaxSingle-shot agent · high reasoningPublished margin ±7.9 · 118 reported samples	91.70%	Not reported	Unknown
44	Sonnet 4.6AnthropicSingle-shot agent · high reasoningPublished margin ±9.2 · 120 reported samples	90.00%	Not reported	Unknown
45	MiniMax-M2.7MiniMaxSingle-shot agent · high reasoningPublished margin ±6.7 · 120 reported samples	88.30%	Not reported	Unknown
46	Sonnet 5AnthropicSingle-shot agent · max reasoningPublished margin ±10.8 · 120 reported samples	88.30%	Not reported	Unknown
47	Gemma 4 31BGoogleSingle-shot agentPublished margin ±11.7 · 120 reported samples	84.20%	Not reported	Unknown
48	Gemini 3.5 FlashGoogleSingle-shot agent · high reasoningPublished margin ±9.6 · 120 reported samples	81.70%	Not reported	Unknown
49	Gemini 3.5 Flash LiteGoogleSingle-shot agent · high reasoningPublished margin ±13.8 · 120 reported samples	70.00%	Not reported	Unknown

How to read this Mercor evaluation

We captured Mercor's AIME 2025 results on October 15, 2026. The table preserves 49 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 = 4. The linked result page takes precedence where its setup differs, as noted above.

Snapshot

Pass@149 configurationsMercor evaluationDisplay only

Mercor AIME 2025 results Mercor APEX benchmark index Mercor methodology Upstream dataset

The published AIME 2025 (Mercor) snapshot places Opus 5 first at 100.00%. The third row is 0.00 points behind. The broader top-10 range is 0.00 points, so many of the published results sit in a relatively narrow band.

49 configurations are shown for AIME 2025 (Mercor). The benchmark falls in the Mathematics category. We keep external benchmark mirrors separate from the weighted global scoring system, so these results remain source-specific evidence. AIME 2025 (Mercor) is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About AIME 2025 (Mercor)

Tasks

30 source-reported tasks

Format

Pass@1

Difficulty

Source-specific evaluation

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.

AIME 2025 evaluation results Public benchmark source

Freshness and provenance

Version

AIME 2025; Mercor Pass@1

Refresh cadence

Reviewed dated source capture

Staleness state

Dated Mercor configurations

Question availability

Upstream task access and terms; only aggregate results mirrored

Dated Mercor configurationsDisplay only

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

Questions

What does AIME 2025 (Mercor) measure?

Solving competition math problems with integer answers. This table shows Mercor-run configurations for reference and is excluded from model rankings.

Which model leads the published AIME 2025 (Mercor) snapshot?

Opus 5 currently leads the published AIME 2025 (Mercor) snapshot with 100.00% pass@1. BenchLM shows this benchmark for display only and does not use it in overall rankings.

How many models are evaluated on AIME 2025 (Mercor)?

The October 15, 2026 capture snapshot contains 49 source configurations.

Last updated: October 15, 2026 capture · mirrored from the public benchmark leaderboard

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 5,600+ readers.

One email each week. Unsubscribe anytime.