Weighted consensus · Public benchmarks

LLM leaderboard

Name: UnifyBench public ranking snapshot
Creator: UnifyBench
Published: 2026-10-31T00:02:06.506Z
License: https://unifybench.ai/data#reuse

Compare AI models across public benchmarks. Source-balanced consensus for overall performance, coding, reasoning and agentic work.

Looking for which LLM is best? UnifyBench is a first stop: source-balanced consensus across many published benchmarks (not one board). Shortlist here, then validate on Artificial Analysis, check Arena for preference, and finish with task-specific evals.

Overall leaders

Comparison score · 0–100

Consensus across available benchmarks. Models cover different task areas; use a task ranking for a focused comparison.

Experimental scores for the selected profile and sources. Small gaps may not be meaningful. How scoring works · Effort evidence collected 2026-10-31

Reviewed retained observations are in use. October 2026 Real-SWE board has no 95% confidence whisker, so these scores stay at the 2026-10-18 retrieval. Original source retrieval 2026-10-18; source updated 2026-10-18.

This month: three boards to compare

Which LLM is best in November 2026?

The dated checks below show different measures of performance: GPT-6 Astra · XHigh is UnifyBench's last known leader; Gemini 4 Carbon (High) leads Artificial Analysis; Gemini 4 Carbon (High) is Arena's last known leader. Shortlist with UnifyBench, check Artificial Analysis for independent scores and cost per task, check Arena for preferences, then test your own tasks.

UnifyBenchGPT-6 Astra · XHigh96.0Overall · benchmark consensus · 0–100Last known leader
Artificial AnalysisGemini 4 Carbon (High)59Intelligence Index · independent suite
ArenaGemini 4 Carbon (High)1528 ± 9Text Overall Elo · human preferenceLast known leaderPreliminary

The Intelligence Index, Arena Elo, and UnifyBench score use different scales and cannot be compared directly. UnifyBench balances many published benchmarks rather than blending a few board ranks. Why leaderboards disagree.

Source dates and refresh details

UnifyBench
Evidence published 31 Oct 2026
Default weights · documented evidence · highest-scoring effort · Earlier-month evidence

Artificial Analysis
Checked 2 Nov 2026 · Source update date unavailable
#1

Arena
Checked 2 Nov 2026 · Board updated 27 Oct 2026
#1 · Preliminary · Earlier-month source

471unique models 569model configurations 135source pages

Explore all 135 source pages Independent evaluations and lab reports

Each model is counted once across effort levels. Configurations count documented model + effort settings, including unranked.

Original reports and benchmark leaderboards. Distinct source pages with published results in the catalog. Multiple results and page sections count as one source.

One entry per model, using its highest-scoring eligible effort setting. How effort ranking works.

Customize ranking Sources, profiles and evidence details

Full ranking

231 entries

Compare two configurations

Capability ranking of measured model configurations. Scores are experimental; missing results remain unknown.
Rank	Model	Score	Evidence breadth Why this score?
126	MiniCPM5-2B · ReasoningOpenBMB	2.8	Single independent source10 families · 4 areas measured
127	GLM-4.7-Flash · ReasoningZ.ai	2.8	Single independent source7 families · 5 areas measured
128	Qwen3-Next-80B-A3B-Thinking · ReasoningQwen	2.7	Single independent source12 families · 5 areas measured
129	GLM-4.5 · ReasoningZ.ai	2.6	Single independent source11 families · 5 areas measured
130	Qwen3-VL-32B-Thinking · ReasoningQwen	2.4	Single independent source11 families · 6 areas measured
131	NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 · ReasoningNVIDIA	2.4	Single independent source13 families · 5 areas measured
132	NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 · ReasoningNVIDIA	2.4	Single independent source14 families · 5 areas measured
133	GPT-5 Nano · HighOpenAI	2.3	Multiple independent sources13 families · 6 areas measured
134	o3-mini · HighOpenAI	2.2	Multiple independent sources15 families · 5 areas measured
135	Magistral Medium 1.2 · ReasoningMistral	2.2	Single independent source13 families · 6 areas measured
136	DiffusionGemma 26B A4B · ReasoningGoogle	2.0	Single independent source8 families · 6 areas measured
137	Seed OSS 36B Instruct · ReasoningByteDance	2.0	Single independent source11 families · 5 areas measured
138	Solar Pro 3 · ReasoningUpstage	1.9	Single independent source12 families · 5 areas measured
139	Qwen3-235B-A22B-Instruct-2507 · Non-reasoningQwen	1.9	Single independent source11 families · 5 areas measured
140	Mistral Small 4 · ReasoningMistral	1.9	Single independent source15 families · 6 areas measured
141	GLM-4.5-Air · ReasoningZ.ai	1.7	Single independent source11 families · 5 areas measured
More models on this page 142 gpt-oss-120b · High 1.7 143 Qwen3.5-omni-flash · Non-reasoning 1.6 144 Qwen3-Coder-Next · Non-reasoning 1.6 145 gpt-oss-20b · High 1.5 146 Qwen3-30B-A3B-Thinking-2507 · Reasoning 1.5 147 MiniMax-M1-80k · Reasoning 1.5 148 Qwen3-VL-235B-A22B-Instruct · Non-reasoning 1.5 149 Qwen3-VL-30B-A3B-Thinking · Reasoning 1.4 150 Solar Open 100B · Reasoning 1.4

Behind the ranking

294 entries with published results · 231 ranked · 7 provisional estimates.

Explore seven capabilities, the original results, and gaps in the evidence.

Coding ranking Reasoning ranking Agentic ranking Knowledge ranking Compare capabilities ↗Published benchmarks ↗Evidence coverage ↗

Cite this ranking

Include the model and effort setting, the selected profile, and the dated data downloads. A shared link uses the latest data; it does not freeze a ranking in time.

UnifyBench ranking
Data updated: 31 Oct 2026
Mode: overall; evidence: documented; configurations: best
https://unifybench.ai/?page=6

Catalog data · Effort measurements · Configuration scoring rules · Methodology

Catalog dated 31 Oct 2026. Effort evidence collected 31 Oct 2026. Collection dates are not evaluation dates.

What does the score mean?

The adjusted comparison score estimates performance against the same reference panel from matched published comparisons. Capability pools matched comparisons across families and checks aggregate breadth and independent source coverage. Custom profiles require support in every selected capability. Missing results never count as losses.

Family details show observed win shares against available reference peers. Those descriptive values are supporting evidence, not adjusted capability scores. Provider reports and independent results remain labeled. Matching reported settings do not establish equal inference effort, and score differences do not establish statistical significance.

Read the full methodology →