Benchmark Almanac

OPT-BENCH

DeepSeek-V3.1 0.69 leads since August 2025
Not in model cards

Agents iteratively improve solutions to 20 machine-learning tasks and 10 combinatorial optimization problems using execution feedback. Win counts, invalid-solution rates, ranks, and improvement rates measure optimization performance.

Results depend on the supplied OPT-Agent refinement protocol and the number of iterations.

Released
June 2025
Kind
Scientific code
Level
University
Scored
Checked automatically
Answers
Code
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Code, Tables → Code, Actions
Interaction
Agentic
Environment
Code sandbox
Refresh
One fixed set
Run by
Anyone
In views
Agents
Strongest model tested
Qwen3-235B-A22B-Thinking Jul 2025 #108 on the capability index
Last model added
August 2025
Scores refreshed
November 20th, 2026

In the authors’ words

Large Language Models (LLMs) have shown remarkable capabilities in solving diverse tasks. However, their proficiency in iteratively optimizing complex solutions through learning from previous feedback remains insufficiently explored. To bridge this gap, we present OPT-BENCH, a comprehensive benchmark designed to evaluate LLM agents on large-scale search space optimization problems. OPT-BENCH includes 20 real-world machine learning tasks sourced from Kaggle and 10 classical NP problems, offering a diverse and challenging environment for assessing LLM agents on iterative reasoning and solution refinement. To enable rigorous evaluation, we introduce OPT-Agent, an end-to-end optimization framework that emulates human reasoning when tackling complex problems by generating, validating, and iteratively improving solutions through leveraging historical feedback. Through extensive experiments on 9 state-of-the-art LLMs from 6 model families, we analyze the effects of optimization iterations, temperature settings, and model architectures on solution quality and convergence. Our results demonstrate that incorporating historical context significantly enhances optimization performance across both ML and NP tasks. All datasets, code, and evaluation tools are open-sourced to promote further research in advancing LLM-driven optimization and iterative reasoning. Project page: https://github.com/OliverLeeXZ/OPT-BENCH.

Abstract of OPT-BENCH: Evaluating LLM Agent on Large-Scale Search Spaces Optimization Problems (arXiv:2506.10764, CC0)
19
Entries in its paper, 15 models we track

DeepSeek-V3.1 leads with 0.69 among models released by August 2025 DeepSeek-V3.1 leads with 0.69 among models released by August 2025

Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. The paper reports no overall score, so the headline is the plain mean of Expert Gap on the ML and NP tracks. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.

The newest model with a score here came out in August 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 DeepSeek-V3.1 20 steps 0.69
  2. 2 o3-mini 20 steps 0.69
  3. 3 Qwen3-235B-A22B-Thinking Jul 2025, 20 steps 0.69
  4. 4 Grok 3 20 steps 0.63
  5. 5 Qwen3-30B-A3B 20 steps 0.61
  6. 6 Claude 3.7 Sonnet 20 steps 0.59
  7. 7 Qwen3-32B 20 steps 0.59
  8. 8 Qwen3-235B-A22B-Instruct Jul 2025, 20 steps 0.58
  9. 9 GPT-4.1 20 steps 0.56
  10. 10 GPT-4o Aug 2024, 20 steps 0.54
  11. 11 Qwen3-8B 20 steps 0.51
  12. 12 Claude 3.5 Sonnet 20 steps 0.50
  13. 13 Gemini 2.0 Flash Feb 2025, 20 steps 0.49
  14. 14 Qwen2.5-32B-Instruct 20 steps 0.48
  15. 15 Qwen2.5-72B 20 steps 0.46
  16. 16 Qwen2.5-14B-Instruct 20 steps 0.39
  17. 17 Qwen2.5-7B-Instruct 20 steps 0.28
  18. 18 InternLM3-8B-Instruct 20 steps 0.27
  19. 19 Qwen2.5-3B-Instruct 20 steps 0.18

Source: The OPT-BENCH paper

#EntryScoreSettingDateSource
1 DeepSeek-V3.1 0.69 20 steps, OPT-Agent Aug 2025 Official leaderboard
2 o3-mini 0.69 20 steps, OPT-Agent Jan 2025 Official leaderboard
3 Qwen3-235B-A22B-Thinking Jul 2025 0.69 20 steps, OPT-Agent Jul 2025 Official leaderboard
4 Grok 3 0.63 20 steps, OPT-Agent Apr 2025 Official leaderboard
5 Qwen3-30B-A3B 0.61 20 steps, OPT-Agent Apr 2025 Official leaderboard
6 Claude 3.7 Sonnet 0.59 20 steps, OPT-Agent Feb 2025 Official leaderboard
7 Qwen3-32B 0.59 20 steps, OPT-Agent Apr 2025 Official leaderboard
8 Qwen3-235B-A22B-Instruct Jul 2025 0.58 20 steps, OPT-Agent Jul 2025 Official leaderboard
9 GPT-4.1 0.56 20 steps, OPT-Agent Apr 2025 Official leaderboard
10 GPT-4o Aug 2024 0.54 20 steps, OPT-Agent Aug 2024 Official leaderboard
11 Qwen3-8B 0.51 20 steps, OPT-Agent Apr 2025 Official leaderboard
12 Claude 3.5 Sonnet 0.50 20 steps, OPT-Agent Jun 2024 Official leaderboard
13 Gemini 2.0 Flash Feb 2025 0.49 20 steps, OPT-Agent Feb 2025 Official leaderboard
14 Qwen2.5-32B-Instruct 0.48 20 steps, OPT-Agent Official leaderboard
15 Qwen2.5-72B 0.46 20 steps, OPT-Agent Sep 2024 Official leaderboard
16 Qwen2.5-14B-Instruct 0.39 20 steps, OPT-Agent Official leaderboard
17 Qwen2.5-7B-Instruct 0.28 20 steps, OPT-Agent Official leaderboard
18 InternLM3-8B-Instruct 0.27 20 steps, OPT-Agent Jan 2025 Official leaderboard
19 Qwen2.5-3B-Instruct 0.18 20 steps, OPT-Agent Official leaderboard
#ModelScoreSettingDateSource
1 DeepSeek-V3.1 0.69 20 steps, OPT-Agent Aug 2025 Official leaderboard
2 o3-mini 0.69 20 steps, OPT-Agent Jan 2025 Official leaderboard
3 Qwen3-235B-A22B-Thinking Jul 2025 0.69 20 steps, OPT-Agent Jul 2025 Official leaderboard
4 Grok 3 0.63 20 steps, OPT-Agent Apr 2025 Official leaderboard
5 Qwen3-30B-A3B 0.61 20 steps, OPT-Agent Apr 2025 Official leaderboard
6 Claude 3.7 Sonnet 0.59 20 steps, OPT-Agent Feb 2025 Official leaderboard
7 Qwen3-32B 0.59 20 steps, OPT-Agent Apr 2025 Official leaderboard
8 Qwen3-235B-A22B-Instruct Jul 2025 0.58 20 steps, OPT-Agent Jul 2025 Official leaderboard
9 GPT-4.1 0.56 20 steps, OPT-Agent Apr 2025 Official leaderboard
10 GPT-4o Aug 2024 0.54 20 steps, OPT-Agent Aug 2024 Official leaderboard
11 Qwen3-8B 0.51 20 steps, OPT-Agent Apr 2025 Official leaderboard
12 Claude 3.5 Sonnet 0.50 20 steps, OPT-Agent Jun 2024 Official leaderboard
13 Gemini 2.0 Flash Feb 2025 0.49 20 steps, OPT-Agent Feb 2025 Official leaderboard
14 Qwen2.5-72B 0.46 20 steps, OPT-Agent Sep 2024 Official leaderboard
15 InternLM3-8B-Instruct 0.27 20 steps, OPT-Agent Jan 2025 Official leaderboard

Rankings by part

We read 2 parts from the benchmark’s paper: OPT-BENCH-ML Expert Gap and OPT-BENCH-NP Expert Gap. Each ranks models on one part of the test. o3-mini leads OPT-BENCH-ML Expert Gap and DeepSeek-V3.1 leads OPT-BENCH-NP Expert Gap. Arrows show how far a model moved from its overall rank among the models we track.

OPT-BENCH-ML Expert Gap

15 models we track have a score here.

  1. 1 o3-mini ↑1 0.65
  2. 2 Grok 3 ↑2 0.63
  3. 3 Qwen3-235B-A22B-Thinking Jul 2025 0.62
  4. 4 GPT-4o Aug 2024 ↑6 0.61
  5. 5 Qwen3-235B-A22B-Instruct Jul 2025 ↑3 0.60
  6. 6 DeepSeek-V3.1 ↓5 0.60
  7. 7 GPT-4.1 ↑2 0.58
  8. 8 Qwen3-32B ↓1 0.58
  9. 9 Qwen3-30B-A3B ↓4 0.54
  10. 10 Gemini 2.0 Flash Feb 2025 ↑3 0.53

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
DeepSeek-V3.1 OPT-Agent
5 steps 10 steps
0.69
o3-mini OPT-Agent
5 steps 10 steps
0.69
Qwen3-235B-A22B-Thinking OPT-Agent
5 steps 20 steps
0.69
Grok 3 OPT-Agent
5 steps 10 steps 20 steps
0.63
Qwen3-30B-A3B OPT-Agent
5 steps 10 steps
0.61
Claude 3.7 Sonnet OPT-Agent
5 steps 10 steps 20 steps
0.59
Qwen3-32B OPT-Agent
5 steps 10 steps 20 steps
0.59
Qwen3-235B-A22B-Instruct OPT-Agent
5 steps 10 steps 20 steps
0.58
GPT-4.1 OPT-Agent
5 steps 10 steps 20 steps
0.56
GPT-4o OPT-Agent
5 steps 10 steps 20 steps
0.54
Qwen3-8B OPT-Agent
5 steps 10 steps 20 steps
0.51
Claude 3.5 Sonnet OPT-Agent
5 steps 10 steps 20 steps
0.50

Source: The OPT-BENCH paper

ModelEffortHarnessScoreCost per taskOutput tokens
DeepSeek-V3.1 20 steps OPT-Agent 0.69
o3-mini 20 steps OPT-Agent 0.69
Qwen3-235B-A22B-Thinking Jul 2025 20 steps OPT-Agent 0.69
DeepSeek-V3.1 10 steps OPT-Agent 0.68
o3-mini 10 steps OPT-Agent 0.68
Qwen3-235B-A22B-Thinking Jul 2025 10 steps OPT-Agent 0.65
o3-mini 5 steps OPT-Agent 0.64
Qwen3-235B-A22B-Thinking Jul 2025 5 steps OPT-Agent 0.63
DeepSeek-V3.1 5 steps OPT-Agent 0.63
Grok 3 20 steps OPT-Agent 0.63
Qwen3-30B-A3B 20 steps OPT-Agent 0.61
Claude 3.7 Sonnet 20 steps OPT-Agent 0.59
Qwen3-30B-A3B 10 steps OPT-Agent 0.59
Qwen3-32B 20 steps OPT-Agent 0.59
Qwen3-235B-A22B-Instruct Jul 2025 20 steps OPT-Agent 0.58
Claude 3.7 Sonnet 10 steps OPT-Agent 0.56
Grok 3 10 steps OPT-Agent 0.56
GPT-4.1 20 steps OPT-Agent 0.56
GPT-4o Aug 2024 20 steps OPT-Agent 0.54
Qwen3-235B-A22B-Instruct Jul 2025 10 steps OPT-Agent 0.54
Claude 3.7 Sonnet 5 steps OPT-Agent 0.53
Qwen3-30B-A3B 5 steps OPT-Agent 0.53
Qwen3-32B 10 steps OPT-Agent 0.53
GPT-4o Aug 2024 10 steps OPT-Agent 0.51
Grok 3 5 steps OPT-Agent 0.51
Qwen3-8B 20 steps OPT-Agent 0.51
Claude 3.5 Sonnet 20 steps OPT-Agent 0.50
Gemini 2.0 Flash Feb 2025 20 steps OPT-Agent 0.49
GPT-4.1 10 steps OPT-Agent 0.48
Qwen3-235B-A22B-Instruct Jul 2025 5 steps OPT-Agent 0.48
GPT-4o Aug 2024 5 steps OPT-Agent 0.47
Qwen2.5-72B 20 steps OPT-Agent 0.46
Gemini 2.0 Flash Feb 2025 10 steps OPT-Agent 0.46
Claude 3.5 Sonnet 10 steps OPT-Agent 0.46
Qwen3-8B 10 steps OPT-Agent 0.45
Gemini 2.0 Flash Feb 2025 5 steps OPT-Agent 0.42
GPT-4.1 5 steps OPT-Agent 0.42
Qwen2.5-72B 10 steps OPT-Agent 0.41
Claude 3.5 Sonnet 5 steps OPT-Agent 0.40
Qwen3-8B 5 steps OPT-Agent 0.40
Qwen2.5-72B 5 steps OPT-Agent 0.36
Qwen3-32B 5 steps OPT-Agent 0.35
InternLM3-8B-Instruct 20 steps OPT-Agent 0.27
InternLM3-8B-Instruct 10 steps OPT-Agent 0.21
InternLM3-8B-Instruct 5 steps OPT-Agent 0.09

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
DeepSeek-V3.1 OPT-Agent
5 steps 10 steps
0.69
o3-mini OPT-Agent
5 steps 10 steps
0.69
Qwen3-235B-A22B-Thinking OPT-Agent
5 steps 20 steps
0.69
Grok 3 OPT-Agent
5 steps 10 steps 20 steps
0.63
Qwen3-30B-A3B OPT-Agent
5 steps 10 steps
0.61
Claude 3.7 Sonnet OPT-Agent
5 steps 10 steps 20 steps
0.59
Qwen3-32B OPT-Agent
5 steps 10 steps 20 steps
0.59
Qwen3-235B-A22B-Instruct OPT-Agent
5 steps 10 steps 20 steps
0.58
GPT-4.1 OPT-Agent
5 steps 10 steps 20 steps
0.56
GPT-4o OPT-Agent
5 steps 10 steps 20 steps
0.54
Qwen3-8B OPT-Agent
5 steps 10 steps 20 steps
0.51
Claude 3.5 Sonnet OPT-Agent
5 steps 10 steps 20 steps
0.50

Source: The OPT-BENCH paper

ModelEffortHarnessScoreCost per taskOutput tokens
DeepSeek-V3.1 20 steps OPT-Agent 0.69
o3-mini 20 steps OPT-Agent 0.69
Qwen3-235B-A22B-Thinking Jul 2025 20 steps OPT-Agent 0.69
DeepSeek-V3.1 10 steps OPT-Agent 0.68
o3-mini 10 steps OPT-Agent 0.68
Qwen3-235B-A22B-Thinking Jul 2025 10 steps OPT-Agent 0.65
o3-mini 5 steps OPT-Agent 0.64
Qwen3-235B-A22B-Thinking Jul 2025 5 steps OPT-Agent 0.63
DeepSeek-V3.1 5 steps OPT-Agent 0.63
Grok 3 20 steps OPT-Agent 0.63
Qwen3-30B-A3B 20 steps OPT-Agent 0.61
Claude 3.7 Sonnet 20 steps OPT-Agent 0.59
Qwen3-30B-A3B 10 steps OPT-Agent 0.59
Qwen3-32B 20 steps OPT-Agent 0.59
Qwen3-235B-A22B-Instruct Jul 2025 20 steps OPT-Agent 0.58
Claude 3.7 Sonnet 10 steps OPT-Agent 0.56
Grok 3 10 steps OPT-Agent 0.56
GPT-4.1 20 steps OPT-Agent 0.56
GPT-4o Aug 2024 20 steps OPT-Agent 0.54
Qwen3-235B-A22B-Instruct Jul 2025 10 steps OPT-Agent 0.54
Claude 3.7 Sonnet 5 steps OPT-Agent 0.53
Qwen3-30B-A3B 5 steps OPT-Agent 0.53
Qwen3-32B 10 steps OPT-Agent 0.53
GPT-4o Aug 2024 10 steps OPT-Agent 0.51
Grok 3 5 steps OPT-Agent 0.51
Qwen3-8B 20 steps OPT-Agent 0.51
Claude 3.5 Sonnet 20 steps OPT-Agent 0.50
Gemini 2.0 Flash Feb 2025 20 steps OPT-Agent 0.49
GPT-4.1 10 steps OPT-Agent 0.48
Qwen3-235B-A22B-Instruct Jul 2025 5 steps OPT-Agent 0.48
GPT-4o Aug 2024 5 steps OPT-Agent 0.47
Qwen2.5-72B 20 steps OPT-Agent 0.46
Gemini 2.0 Flash Feb 2025 10 steps OPT-Agent 0.46
Claude 3.5 Sonnet 10 steps OPT-Agent 0.46
Qwen3-8B 10 steps OPT-Agent 0.45
Gemini 2.0 Flash Feb 2025 5 steps OPT-Agent 0.42
GPT-4.1 5 steps OPT-Agent 0.42
Qwen2.5-72B 10 steps OPT-Agent 0.41
Claude 3.5 Sonnet 5 steps OPT-Agent 0.40
Qwen3-8B 5 steps OPT-Agent 0.40
Qwen2.5-72B 5 steps OPT-Agent 0.36
Qwen3-32B 5 steps OPT-Agent 0.35
InternLM3-8B-Instruct 20 steps OPT-Agent 0.27
InternLM3-8B-Instruct 10 steps OPT-Agent 0.21
InternLM3-8B-Instruct 5 steps OPT-Agent 0.09

2 models have held number one in the benchmark’s paper since its release in June 2025

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI DeepSeek
Benchmark paper Not tested since Aug 2025

Source: The OPT-BENCH paper

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
In the benchmark’s paper
0.00 0.20 0.40 0.60 0.80 2024 2025 0.00 0.20 0.40 0.60 0.80 2024 2025

Source: Official OPT-BENCH leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Coding

All 523 in Coding
  • SWE-bench Verified In use 500 human-validated GitHub issues from 12 Python repos. Agent writes a patch that must pass hidden tests. Claude Opus 4.7 83.5%
  • LiveCodeBench v6 In use Competitive programming problems collected after model cutoffs from LeetCode, AtCoder, Codeforces. Scored pass@1 on hidden tests. DeepSeek-V4-Pro 93.5%
  • SWE-bench Multilingual In use 300 GitHub-issue tasks across nine programming languages in SWE-bench format. Patch graded by repository tests. Claude Fable 5.2 96.6%
  • DeepSWE In use 113 original long-horizon engineering tasks across 91 repos and five languages, written from scratch. Hand-written verifiers accept alternative solutions. GPT-6 Astra 74.1%
  • SciCode In use Scientist-curated research coding problems split into subproblems across natural sciences. Code checked against test cases. Seed2.1 Pro 59.8%
  • Codeforces rating In use Model solves recent Codeforces contest problems. An Elo-style rating is computed from simulated contest performance. DeepSeek V4.1 Flash 3,471

Data audit

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 15 months ago, in August 2025, for DeepSeek-V3.1. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
  • Pass Independent runs Scores from the benchmark’s paper (19 entries).
  • Pass Models scored 15 models we track, plus 4 other entries on its leaderboard.
  • Warning Newest score The newest came 15 months ago, in August 2025, for DeepSeek-V3.1.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Qwen3-235B-A22B-Thinking (#108).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on arxiv.org: 45 rows for models we track and 4 other entries at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check