Noometry

Coding benchmark

ALE-Bench leaderboard

As of October 2026, GPT-6 Astra has the highest published ALE-Bench score on Noometry at 2,951, out of 105 models with results.

Last verified

About ALE-Bench

Hard optimization problems from AtCoder Heuristic Contests, scored with a contest-style performance rating.

Category
Coding
Introduced
2025
Format
Heuristic optimization
Unit
Raw score
Official site
sakanaai.github.io

Top 15 models

  1. GPT-6 Astra 2,951
  2. GPT-6 Sol 2,462
  3. GPT-5.6 Sol 2,177
  4. Claude Opus 5 2,165
  5. Claude Opus 5.5 2,147
  6. Claude Fable 5.1 2,143
  7. Claude Fable 5 2,041
  8. GPT-5.6 Terra 1,951
  9. GPT-5.5 1,943
  10. Claude Sonnet 5.5 1,819
  11. GPT-5.6 Luna 1,667
  12. GPT-5.3 Codex 1,655
  13. GPT-5.4 1,607
  14. GPT-6 Luna 1,577
  15. Claude Opus 4.8 1,564

Sponsored placements are available on pages like this one. Advertise on Noometry

All results

#ModelProviderRatingSettingSourceDate
1GPT-6 Astra OpenAI2,951maxEpoch AI
2GPT-6 Sol OpenAI2,462maxEpoch AI
3GPT-5.6 Sol OpenAI2,177maxEpoch AI
4Claude Opus 5 Anthropic2,165highEpoch AI
5Claude Opus 5.5 Anthropic2,147highEpoch AI
6Claude Fable 5.1 Anthropic2,143highEpoch AI
7Claude Fable 5 Anthropic2,041highEpoch AI
8GPT-5.6 Terra OpenAI1,951maxEpoch AI
9GPT-5.5 OpenAI1,943xhighEpoch AI
10Claude Sonnet 5.5 Anthropic1,819highEpoch AI
11GPT-5.6 Luna OpenAI1,667maxEpoch AI
12GPT-5.3 Codex OpenAI1,655xhighEpoch AI
13GPT-5.4 OpenAI1,607highEpoch AI
14GPT-6 Luna OpenAI1,577xhighEpoch AI
15Claude Opus 4.8 Anthropic1,564highEpoch AI
16Kimi K3 Moonshot AI1,524maxEpoch AI
17Grok 4.6 xAI1,508xhighEpoch AI
18Claude Sonnet 5 Anthropic1,463highEpoch AI
19DeepSeek V4 Pro DeepSeek1,403maxEpoch AI
20Gemini 3 Flash Preview Google1,367Epoch AI
21Claude Sonnet 4.6 Anthropic1,327mediumEpoch AI
22Claude Opus 4.7 Anthropic1,323Epoch AI
23GLM-5.3 Z.ai (Zhipu)1,317highEpoch AI
24Grok 4.5 xAI1,309highEpoch AI
25DeepSeek V4 Flash DeepSeek1,306maxEpoch AI
26GPT-5.2 Codex OpenAI1,300Epoch AI
27GPT-5.2 OpenAI1,294highEpoch AI
28Gemini 3.8 Flash Google1,270highEpoch AI
29GPT-5.1-Codex OpenAI1,245Epoch AI
30GPT-5.1 OpenAI1,192highEpoch AI
31Qwen3.7 Max Alibaba (Qwen)1,189Epoch AI
32GPT-5.4 mini OpenAI1,189highEpoch AI
33Gemini 3 Pro Google1,177Epoch AI
34GPT-5 OpenAI1,162highEpoch AI
35Gemini 3.1 Pro Preview Google1,161Epoch AI
36MiMo-V2.6-Pro Xiaomi1,158Epoch AI
37Grok 4.20 (Non-Reasoning) xAI1,150Epoch AI
38Kimi K2.6 Moonshot AI1,093Epoch AI
39DeepSeek V4.1 Flash DeepSeek1,092maxEpoch AI
40GLM-5.2 Z.ai (Zhipu)1,047highEpoch AI
41Claude Opus 4.5 Anthropic1,02516KEpoch AI
42GPT-5.4 nano OpenAI1,005highEpoch AI
43Claude Opus 4.6 Anthropic996.5Epoch AI
44Inkling Thinking Machines Lab946Epoch AI
45Grok 4.3 xAI944.17Epoch AI
46o3 OpenAI933.55highEpoch AI
47Gemma 4 26B A4B IT Google927.17Epoch AI
48Gemma 4 31B IT Google925.5Epoch AI
49Gemini 3.5 Flash Google911.02highEpoch AI
50Gemini 3.7 Flash Google904.3highEpoch AI
51MiMo-V2.5-Pro Xiaomi899.8Epoch AI
52GLM-5.1 Z.ai (Zhipu)887.1Epoch AI
53Kimi K2.7 Code Moonshot AI886.23Epoch AI
54o4-mini OpenAI826.17highEpoch AI
55Kimi K2.5 Moonshot AI821.65Epoch AI
56DeepSeek-R1 DeepSeek804.12Epoch AI
57GPT-5 Mini OpenAI799.77highEpoch AI
58Gemini 3.1 Flash Lite Google797.73Epoch AI
59Claude Sonnet 4.5 Anthropic796.1532KEpoch AI
60Mercury 2 Inception785.58Epoch AI
61Gemini 2.5 Pro Google785.5232KEpoch AI
62MiMo-V2-Pro Xiaomi785.17Epoch AI
63GLM-5 Z.ai (Zhipu)765.62Epoch AI
64Gemini 3.5 Flash Lite Google765.27highEpoch AI
65Mistral Medium Mistral AI763.98Epoch AI
66DeepSeek-V3.1-Terminus DeepSeek745.17Epoch AI
67MiMo-V2-Flash Xiaomi737.95Epoch AI
68GPT-5 Nano OpenAI718.67highEpoch AI
69Gemini 3.6 Flash Google715.52highEpoch AI
70Step 3.7 Flash StepFun694.12Epoch AI
71Claude Opus 4.1 Anthropic674.7716KEpoch AI
72Qwen3.6 Plus Alibaba (Qwen)670.15Epoch AI
73Gemini 2.5 Flash Google661.88Epoch AI
74Claude Sonnet 4 Anthropic655.3532KEpoch AI
75Claude Haiku 4.5 Anthropic653.4832KEpoch AI
76MiniMax-M3 MiniMax640.02Epoch AI
77MiniMax-M2.1 MiniMax623.83Epoch AI
78Qwen3.5 Plus Alibaba (Qwen)621.92Epoch AI
79MiniMax-M2.5 MiniMax618.17Epoch AI
80MiniMax-M2.7 MiniMax599.25Epoch AI
81Kimi K2 (Jul 2025) Moonshot AI597.5Epoch AI
82gpt-oss-120b OpenAI575.62Epoch AI
83gpt-oss-20b OpenAI566.05Epoch AI
84GPT-4.1 OpenAI558.1Epoch AI
85MiMo-V2.5 Xiaomi513.95Epoch AI
86Mistral Small Mistral AI497.62Epoch AI
87Qwen3-Coder 480B-A35B Instruct Alibaba (Qwen)461.45Epoch AI
88Qwen3 Coder Plus Alibaba (Qwen)456.5Epoch AI
89Ring-2.6-1T Ant Group (inclusionAI)432.57Epoch AI
90GLM-4.7 Z.ai (Zhipu)399.48Epoch AI
91Grok 4.1 Fast xAI394.93Epoch AI
92Qwen3 Max Alibaba (Qwen)370.45Epoch AI
93Qwen3.5 27B Alibaba (Qwen)349.45Epoch AI
94GLM-4.5 Z.ai (Zhipu)344.82Epoch AI
95GLM-4.6 Z.ai (Zhipu)340.82Epoch AI
96Qwen3.6 Flash Alibaba (Qwen)326.4Epoch AI
97Gemini 2.5 Flash-Lite Google325.9Epoch AI
98GLM-5.3-Flash Z.ai (Zhipu)303.55highEpoch AI
99Mercury 2.5 Inception301.65highEpoch AI
100Mistral Large Mistral AI264.7Epoch AI
101Amazon Nova Lite Amazon236.25Epoch AI
102Qwen3.5-Flash Alibaba (Qwen)221.8Epoch AI
103Nemotron 3 Super NVIDIA213.9Epoch AI
104Llama 4 Maverick Meta172.97Epoch AI
105Codestral Mistral AI137.78Epoch AI

Compare the leaders

Other coding benchmarks

Frequently asked questions

What does ALE-Bench measure?

Hard optimization problems from AtCoder Heuristic Contests, scored with a contest-style performance rating.

Which model has the highest ALE-Bench score?

As of October 2026, GPT-6 Astra has the highest published ALE-Bench score on Noometry at 2,951, out of 105 models with results.

What is the best open-weight model on ALE-Bench?

Kimi K3 has the highest ALE-Bench score among open-weight models at 1,524, ranking 16 of 105 overall.