CANONICAL BENCHMARK • EVALUATION SPECIFICATION

LiveCodeBench Leaderboard: Contamination-Free Coding (2026)

Contamination-free competitive programming problems from LeetCode, Codeforces, and AtCoder updated continuously.

What LiveCodeBench Leaderboard: Contamination-Free Coding (2026) Actually Measures

Algorithmic problem solving, complex mathematical reasoning, and edge-case handling on brand-new competitive programming problems published after model training cutoff dates.

⚠ Known Boundaries & Harness Sensitivity

Measures competitive programming and algorithmic puzzles rather than real-world software maintenance or multi-file application development.

Intelligence Index vs. Cost per 1M Tokens

Code Intelligence Index vs. Weighted average cost (USD, Log Scale)

Most attractive quadrant Pareto frontier
Direct Answer • What Is LiveCodeBench?

LiveCodeBench evaluates large language models on continuously updated competitive programming problems published after each model’s training knowledge cutoff date. By sourcing fresh contest challenges from LeetCode, AtCoder, and Codeforces, it provides a leak-free Pass@1 measurement of raw algorithmic reasoning.

LiveCodeBench Leaderboard top 10 models ranked by contamination-free Pass@1 accuracy
Figure 1: LiveCodeBench Top 10 Models on post-training cutoff contest problems (Verified October 18, 2026).
### What LiveCodeBench Measures LiveCodeBench tackles the severe problem of pre-training benchmark data contamination. Many LLMs score artificially high on older benchmarks (like HumanEval and MBPP) because test solutions were included in their web crawl training sets. LiveCodeBench continuously collects new, time-stamped contest problems from platforms like LeetCode, Codeforces, and AtCoder that post-date the training cutoff of all tested models.
Page updated: 400 Evaluated Tasks • Metric: Pass@1
SOURCE & HARNESS
#01 Claude Opus 5.5 Anthropic 51.2% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#02 GPT-6 Astra (max) OpenAI 50.5% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#03 GPT-6 Astra (max) OpenAI 50.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#04 Claude Fable 5.1 (max with fallback) Anthropic 50.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#05 Claude Fable 5.1 (max with fallback) Anthropic 50.0% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#06 Muse Wonder Meta 49.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#07 Claude Opus 5 (max) Benchmarks, Price & Specs (Oct 2026) Anthropic 48.0% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#08 Claude Opus 5 (max) Benchmarks, Price & Specs (Oct 2026) Anthropic 48.0% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported
#09 Gemini 4 Argon (high) Google 47.5% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#10 Claude Sonnet 5.5 Anthropic 47.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#11 GPT-6 Sol OpenAI 45.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#12 Muse Spark 1.3 (max) Meta 45.0% — 1 Sep 2026 Pc_direct_sync →
Llama-3.1-Community reported standard
#13 Muse Spark 1.3 (max) Meta 45.0% — 1 Sep 2026 Pc_direct_sync →
Llama-3.1-Community reported
#14 GPT-5.6 Sol (max) Benchmarks, Price & Specs (Oct 2026) OpenAI 44.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#15 GPT-5.6 Sol (max) Benchmarks, Price & Specs (Oct 2026) OpenAI 44.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported
#16 Mistral Large 4 (preview) Mistral 43.5% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#17 Qwen3.8 Max (0902) Alibaba 43.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#18 Kimi K3 (max) Kimi 42.5% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#19 GLM-5.3 (max) Z AI 42.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#20 GLM-5.3 (max) Z AI 42.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported
#21 Grok 4.7 SpaceXAI 42.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#22 Step 5 Preview (preview) StepFun 42.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#23 Qwen3.8 Max (0902) Alibaba 42.0% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#24 Kimi K3 (max) Kimi 41.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#25 Step 5 Preview (preview) StepFun 41.2% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#26 Muse Spark Open (816B MoE) Meta 41.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#27 Grok 4.6 (high) SpaceXAI 41.0% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#28 Grok 4.6 (high) SpaceXAI 41.0% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported
#29 DeepSeek V4.1 Flash (1017 checkpoint) DeepSeek 40.5% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#30 GLM-5.3-Flash Z AI 39.8% — 1 Sep 2026 Pc_direct_sync →
Apache-2.0 reported standard
#31 GLM-5.3-Flash Z AI 39.8% — 1 Sep 2026 Pc_direct_sync →
Apache-2.0 reported
#32 Gemini 3.8 Flash (high) Google 39.5% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#33 GPT-5.6 Terra (max) OpenAI 39.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#34 GPT-5.6 Terra (max) OpenAI 39.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported
#35 Qwen3.8-Flash-Next Alibaba 38.5% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#36 Gemini 3.8 Flash (high) Google 38.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#37 GLM-5.3 Fast (highspeed) Z AI 38.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#38 Qwen3.8-Flash-Next Alibaba 38.0% — 1 Sep 2026 Pc_direct_sync →
Apache-2.0 reported standard
#39 DeepSeek V4.1 Flash (max) DeepSeek 37.8% — 1 Sep 2026 Pc_direct_sync →
MIT reported
#40 DeepSeek V4.1 Flash (max) DeepSeek 37.8% — 1 Sep 2026 Pc_direct_sync →
MIT reported standard
#41 GPT-6 Luna OpenAI 37.5% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#42 Qwen3.8 2.4T A95B Alibaba 37.5% — 1 Sep 2026 Pc_direct_sync →
Apache-2.0 reported standard
#43 Qwen3.8 2.4T A95B Alibaba 37.5% — 1 Sep 2026 Pc_direct_sync →
Apache-2.0 reported
#44 Claude Haiku 5.5 Anthropic 37.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#45 Claude Sonnet 5 (max) Anthropic 36.0% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported
#46 Claude Sonnet 5 (max) Anthropic 36.0% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#47 MiniMax M3.1 Flash (max) MiniMax 35.5% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#48 GPT-5.6 Luna (max) OpenAI 35.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported
#49 GPT-5.6 Luna (max) OpenAI 35.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#50 DeepSeek V4 Pro (0813 max) DeepSeek 35.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
Sources & Open Data Licences CC-BY / Open Source

In compliance with upstream dataset licenses, all evaluated telemetry rows are credited to their primary sources:

  • Manual / Direct Evaluation Ingest — Licence: Editorial
Sources & Open Data Licences CC-BY / Open Source

In compliance with upstream dataset licenses, all evaluated telemetry rows are credited to their primary sources:

What it does not measure

Every benchmark is narrow by design. Reading LiveCodeBench as a general statement about coding ability is the most common mistake people make with it. Here is what the number is silent on:

  • Software engineering. Contest problems are algorithmic puzzles with a single correct output, which is almost the opposite of production work.
  • Working with existing code. Every problem starts from a blank file.
  • Maintainability, since nothing is maintained.

How to read a score honestly

  • Always read the date window. LiveCodeBench exists to combat training-data contamination by using problems published after a model’s cutoff. A score is only meaningful alongside the window it was measured over.
  • Compare like windows. Two models evaluated over different date ranges faced different problems.
  • Expect divergence from agentic benchmarks. Models that excel at contest problems often place mid-table on SWE-bench, and the reverse is common too.

The trap

A high LiveCodeBench score means a model is good at competitive programming. It is weak evidence that the model will be useful in your repository.

Why we publish this page

Benchmark figures circulate faster than they are checked. Numbers get copied between articles until the original source is several links away and nobody has opened the leaderboard. We maintain this page so there is a version of the figure that names its source, its date and its methodology — and so that when a number here changes, you can see what changed.

Last reviewed 7 September 2026. If a figure on this page disagrees with the official leaderboard, that is a mistake on our side and we want to know about it.

You made it to the end!

Get the weekly briefing

New model prices, benchmark results and Claude Code tips. One email a week. Free.

No spam. Unsubscribe anytime. Privacy policy

Keep exploring

ComparisonKimi K2.5 vs Qwen 3.5: Which Open Model Is Better for Coding, Agents and Running Locally?ReviewWinston AI Detector Review 2026: Is It Accurate? We Ran 4 TestsGuideWhat Is Vibe Coding? Definition, Meaning and How It Works

Complete Verified Leaderboard

Fully server-side rendered table with verified primary source attributions and direct repository links.

Page updated: 400 Evaluated Tasks • Metric: Pass@1
SOURCE & HARNESS
#01 Claude Opus 5.5 Anthropic 51.2% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#02 GPT-6 Astra (max) OpenAI 50.5% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#03 GPT-6 Astra (max) OpenAI 50.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#04 Claude Fable 5.1 (max with fallback) Anthropic 50.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#05 Claude Fable 5.1 (max with fallback) Anthropic 50.0% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#06 Muse Wonder Meta 49.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#07 Claude Opus 5 (max) Benchmarks, Price & Specs (Oct 2026) Anthropic 48.0% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#08 Claude Opus 5 (max) Benchmarks, Price & Specs (Oct 2026) Anthropic 48.0% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported
#09 Gemini 4 Argon (high) Google 47.5% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#10 Claude Sonnet 5.5 Anthropic 47.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#11 GPT-6 Sol OpenAI 45.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#12 Muse Spark 1.3 (max) Meta 45.0% — 1 Sep 2026 Pc_direct_sync →
Llama-3.1-Community reported standard
#13 Muse Spark 1.3 (max) Meta 45.0% — 1 Sep 2026 Pc_direct_sync →
Llama-3.1-Community reported
#14 GPT-5.6 Sol (max) Benchmarks, Price & Specs (Oct 2026) OpenAI 44.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#15 GPT-5.6 Sol (max) Benchmarks, Price & Specs (Oct 2026) OpenAI 44.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported
#16 Mistral Large 4 (preview) Mistral 43.5% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#17 Qwen3.8 Max (0902) Alibaba 43.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#18 Kimi K3 (max) Kimi 42.5% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#19 GLM-5.3 (max) Z AI 42.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#20 GLM-5.3 (max) Z AI 42.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported
#21 Grok 4.7 SpaceXAI 42.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#22 Step 5 Preview (preview) StepFun 42.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#23 Qwen3.8 Max (0902) Alibaba 42.0% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#24 Kimi K3 (max) Kimi 41.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#25 Step 5 Preview (preview) StepFun 41.2% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#26 Muse Spark Open (816B MoE) Meta 41.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#27 Grok 4.6 (high) SpaceXAI 41.0% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#28 Grok 4.6 (high) SpaceXAI 41.0% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported
#29 DeepSeek V4.1 Flash (1017 checkpoint) DeepSeek 40.5% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#30 GLM-5.3-Flash Z AI 39.8% — 1 Sep 2026 Pc_direct_sync →
Apache-2.0 reported standard
#31 GLM-5.3-Flash Z AI 39.8% — 1 Sep 2026 Pc_direct_sync →
Apache-2.0 reported
#32 Gemini 3.8 Flash (high) Google 39.5% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#33 GPT-5.6 Terra (max) OpenAI 39.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#34 GPT-5.6 Terra (max) OpenAI 39.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported
#35 Qwen3.8-Flash-Next Alibaba 38.5% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#36 Gemini 3.8 Flash (high) Google 38.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#37 GLM-5.3 Fast (highspeed) Z AI 38.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#38 Qwen3.8-Flash-Next Alibaba 38.0% — 1 Sep 2026 Pc_direct_sync →
Apache-2.0 reported standard
#39 DeepSeek V4.1 Flash (max) DeepSeek 37.8% — 1 Sep 2026 Pc_direct_sync →
MIT reported
#40 DeepSeek V4.1 Flash (max) DeepSeek 37.8% — 1 Sep 2026 Pc_direct_sync →
MIT reported standard
#41 GPT-6 Luna OpenAI 37.5% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#42 Qwen3.8 2.4T A95B Alibaba 37.5% — 1 Sep 2026 Pc_direct_sync →
Apache-2.0 reported standard
#43 Qwen3.8 2.4T A95B Alibaba 37.5% — 1 Sep 2026 Pc_direct_sync →
Apache-2.0 reported
#44 Claude Haiku 5.5 Anthropic 37.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#45 Claude Sonnet 5 (max) Anthropic 36.0% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported
#46 Claude Sonnet 5 (max) Anthropic 36.0% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#47 MiniMax M3.1 Flash (max) MiniMax 35.5% — 21 Oct 2026 Pc_direct_sync →
MIT reported
#48 GPT-5.6 Luna (max) OpenAI 35.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported
#49 GPT-5.6 Luna (max) OpenAI 35.5% — 1 Sep 2026 Pc_direct_sync →
Proprietary reported standard
#50 DeepSeek V4 Pro (0813 max) DeepSeek 35.0% — 21 Oct 2026 Pc_direct_sync →
MIT reported
Sources & Open Data Licences CC-BY / Open Source

In compliance with upstream dataset licenses, all evaluated telemetry rows are credited to their primary sources:

  • Manual / Direct Evaluation Ingest — Licence: Editorial

What This Benchmark Does Not Measure

Does not measure real-world repo navigation or git workflows.

How to Reproduce This Evaluation

All evaluations published by Vibecoder Journal follow publicly documented harness specifications. You can execute this test suite against any local or API model: 

# Standardized containerized evaluation execution
git clone https://livecodebench.github.io/ && cd evaluation
docker build -t vcj-harness .
docker run --rm -v $(pwd)/results:/results vcj-harness --benchmark livecodebench
Sources & Open Data Licences CC-BY / Open Source

In compliance with upstream dataset licenses, all evaluated telemetry rows are credited to their primary sources:

Abdullah Zulfiqar
Abdullah Zulfiqar Founder & Technical EditorLast updated 18 October 2026

Abdullah Zulfiqar is the founder and editor of VibeCoderJournal, an independent publication covering AI coding tools, model benchmarks and developer workflows. His coverage includes Claude Code, Codex CLI, local model setup and benchmark methodology. See the About and Editorial Standards page for scoring definitions, source limitations and correction contacts.

All Benchmarks → LinkedIn Profile →

Related AI Coding Benchmarks

benchmark 2 min read

SciCode

SciCode evaluates code generation for scientific research problems. Scientists curate problems that combine domain knowledge, reasoning and…

By Abdullah Zulfiqar • Oct 27, 2026
benchmark 2 min read

GPQA Diamond

GPQA Diamond tests graduate-level scientific question answering. It is a subset of GPQA, a multiple-choice benchmark covering biology…

By Abdullah Zulfiqar • Oct 27, 2026