THE LEADERBOARD
Numbers on
the board.
Overall and category scores are grounded in current, comparable records. Seven-day change needs daily snapshots.
| # | Model | Quantum Score | Coding | Reasoning | REVAL™ⓘREVAL™ Research Evaluation Index — our proprietary 0–100 benchmark measuring real-world AI research performance. Methodology → | Agents | Speed | Value |
|---|---|---|---|---|---|---|---|---|
| 01 | Claude Opus 5.5Anthropic | 88 | 94 | 91 | Not tested | 96 | 90 tok/sMax · 2026-10-01 | 52 |
| 02 | Claude Sonnet 5.5Anthropic | 84 | 95 | 91 | Not tested | 78 | Not measured | 62 |
| 03 | GPT-6 AstraOpenAI | 74 | 74 | 94 | Not tested | 60 | 52 tok/sMax · 2026-10-01 | 27 |
| 04 | Claude Fable 5.1Anthropic | 68 | 70 | 86 | Not tested | 70 | Not measured | 25 |
| 05 | Gemini 3.7 FlashGoogle | 62 | 47 | 91 | Not tested | 20 | Not measured | 56 |
| 06 | DeepSeek V4.1 FlashDeepSeek | 60 | 46 | 92 | Not tested | 100 | Not measured | 58 |
| 07 | Gemini 3.8 FlashGoogle | 57 | 32 | 61 | Not tested | 30 | Not measured | 54 |
| 08 | GPT-5.6 TerraOpenAI | 54 | 49 | 67 | Not tested | 34 | 93.6 tok/sMax · 2026-10-01 | 40 |
| 09 | Gemini 3.5 FlashGoogle | 53 | 32 | 75 | Not tested | 20 | Not measured | 43 |
| 10 | GPT-6 SolOpenAI | 52 | 47 | 67 | Not tested | 20 | Not measured | 39 |
| 11 | GPT-5.6 SolOpenAI | 52 | 75 | 69 | Not tested | 20 | Not measured | 29 |
| 12 | GPT-5.6 LunaOpenAI | 51 | 61 | 51 | Not tested | 20 | Not measured | 48 |
| 13 | Grok 4.6xAI | 51 | 46 | 67 | Not tested | 39 | Not measured | 41 |
| 14 | Gemini 3.1 Pro PreviewGoogle | 46 | 19 | 81 | Not tested | 0 | Not measured | 35 |
| 15 | Grok 4.5xAI | 45 | 19 | 90 | Not tested | 32 | Not measured | 38 |
| 16 | GPT-6 LunaOpenAI | 41 | 33 | 39 | Not tested | 22 | 122.4 tok/sMax · 2026-10-01 | 41 |
| 17 | GPT-6.1 SolOpenAI | Not qualifiedQualified category evidence: coding, reasoning, agents, multimodal. An overall score is withheld because only 7 independent observations (minimum 8). Evidence still missing or lacking comparable overlap: math, research, speed. See evidence → | 88 | 83 | Not tested | 85 | 64 tok/sMax · 2026-10-01 | 54 |
| 18 | Claude Haiku 4.5Anthropic | Not qualifiedAn overall score is withheld because only 0 benchmark source domains (minimum 3); only 0 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 0 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |
| 19 | Gemini 3.6 FlashGoogle | Not qualifiedQualified category evidence: coding, reasoning, math, agents. An overall score is withheld because only 2 benchmark source domains (minimum 3). Evidence still missing or lacking comparable overlap: research, multimodal, speed. See evidence → | 47 | 79 | Not tested | 20 | 176.6 tok/sHigh · 2026-10-01 | 48 |
| 20 | Gemini 3.5 Flash-LiteGoogle | Not qualifiedQualified category evidence: coding, reasoning, math, agents. An overall score is withheld because only 1 benchmark source domains (minimum 3); only 7 independent observations (minimum 8); only 1 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: research, multimodal, speed. See evidence → | 34 | 0 | Not tested | 0 | 354.5 tok/sProvider default (not specified by source) · 2026-10-01 | 8 |
| 21 | DeepSeek V4 ProDeepSeek | Not qualifiedQualified category evidence: coding, reasoning, math, agents. An overall score is withheld because only 2 benchmark source domains (minimum 3). Evidence still missing or lacking comparable overlap: research, multimodal, speed. See evidence → | 33 | 95 | Not tested | 40 | Not measured | 52 |
| 22 | Mistral Medium 3.5Mistral | Not qualifiedAn overall score is withheld because only 0 benchmark source domains (minimum 3); only 0 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 0 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |
| 23 | Mistral Small 4Mistral | Not qualifiedAn overall score is withheld because only 0 benchmark source domains (minimum 3); only 0 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 0 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |
| 24 | Mistral Large 3Mistral | Not qualifiedAn overall score is withheld because only 0 benchmark source domains (minimum 3); only 0 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 0 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |
| 25 | Llama 4 ScoutMeta | Not qualifiedAn overall score is withheld because only 0 benchmark source domains (minimum 3); only 0 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 0 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |
| 26 | Llama 4 MaverickMeta | Not qualifiedAn overall score is withheld because only 0 benchmark source domains (minimum 3); only 0 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 0 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |
| 27 | Qwen3.5-397B-A17BQwen / Alibaba | Not qualifiedAn overall score is withheld because only 0 benchmark source domains (minimum 3); only 0 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 0 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |
| 28 | Qwen3.5-122B-A10BQwen / Alibaba | Not qualifiedAn overall score is withheld because only 0 benchmark source domains (minimum 3); only 0 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 0 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |
| 29 | gpt-4.1OpenAI | Not qualifiedAn overall score is withheld because only 0 benchmark source domains (minimum 3); only 0 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 0 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |
| 30 | gpt-4.1-miniOpenAI | Not qualifiedAn overall score is withheld because only 0 benchmark source domains (minimum 3); only 0 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 0 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |
| 31 | gpt-4.1-nanoOpenAI | Not qualifiedAn overall score is withheld because only 0 benchmark source domains (minimum 3); only 0 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 0 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |
| 32 | gpt-5OpenAI | Not qualifiedAn overall score is withheld because only 0 benchmark source domains (minimum 3); only 0 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 0 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |
| 33 | gpt-5-miniOpenAI | Not qualifiedAn overall score is withheld because only 0 benchmark source domains (minimum 3); only 0 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 0 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |
| 34 | gpt-5-nanoOpenAI | Not qualifiedAn overall score is withheld because only 0 benchmark source domains (minimum 3); only 0 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 0 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |
| 35 | gpt-5.1OpenAI | Not qualifiedAn overall score is withheld because only 0 benchmark source domains (minimum 3); only 0 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 0 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |
| 36 | gpt-5.2OpenAI | Not qualifiedAn overall score is withheld because only 0 benchmark source domains (minimum 3); only 0 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 0 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |
| 37 | gpt-5.4-miniOpenAI | Not qualifiedAn overall score is withheld because only 0 benchmark source domains (minimum 3); only 0 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 0 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |
| 38 | gpt-5.4-nanoOpenAI | Not qualifiedAn overall score is withheld because only 0 benchmark source domains (minimum 3); only 0 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 0 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |
| 39 | Gemini 4 ArgonGoogle | Not qualifiedQualified category evidence: coding, reasoning, agents. An overall score is withheld because only 2 benchmark source domains (minimum 3); only 7 independent observations (minimum 8); only 49% of the required weighted category coverage; only 3 qualified categories (minimum 4). Evidence still missing or lacking comparable overlap: math, research, multimodal, speed. See evidence → | 48 | 100 | Not tested | 73 | Not measured | No data |
| 40 | Muse WonderMeta | Not qualifiedAn overall score is withheld because only 1 benchmark source domains (minimum 3); only 4 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 1 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |
| 41 | Muse Wonder FullMeta | Not qualifiedAn overall score is withheld because only 0 benchmark source domains (minimum 3); only 0 independent observations (minimum 8); only 0% of the required weighted category coverage; only 0 qualified categories (minimum 4); only 0 independent source domains (minimum 2). Evidence still missing or lacking comparable overlap: coding, reasoning, math, research, agents, multimodal, speed. See evidence → | No data | No data | Not tested | No data | Not measured | No data |