UnifyBench

Weighted consensus · Public benchmarks

LLM leaderboard

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.

See the full ranking

Capability, within your budget. Compare the same scores using a shared cost basis.

Token counts for price scenarios

Output includes reasoning. Fixed token counts compare prices; they do not measure task efficiency.

Cost coverage and methodology

Capability scores stay unchanged. Lower cost breaks exact score ties; no score-to-price ratio is used. Small score gaps do not establish superiority.

0 of 444 supported configurations have comparable costs. Unknown costs and over-budget entries remain visible without a Value rank. Prices collected 2026-09-22; comparisons expire after 30 days.

USD per weighted AA v4.3 task, including reasoning, cache treatment and reviewed fallback routing. This workload does not represent every task. Cost evidence currently comes from Artificial Analysis only. Cost frontier, components and gaps ↗ · Value methodology

The price collection needs refreshing. Value ranks are suspended; Capability remains available.

Value leaders

Comparison score · 0–100

    No supported leaders match these filters. Clear the search or explore the results below.

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

    This month: three boards to compare

    Which LLM is best in October 2026?

    The dated checks below show different measures of performance: GPT-6 Astra · XHigh leads UnifyBench; Claude Opus 5.5 (Max, Default Fallback) leads Artificial Analysis; Gemini 4 Argon · 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 · XHigh95.9Overall · benchmark consensus · 0–100
    • Artificial AnalysisClaude Opus 5.5 (Max, Default Fallback)58Intelligence Index · independent suite
    • ArenaGemini 4 Argon · High1525 ± 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 22 Oct 2026
    Default weights · documented evidence · highest-scoring effort

    Artificial Analysis
    Checked 28 Oct 2026 · Source update date unavailable
    #1

    Arena
    Checked 7 Oct 2026 · Board updated 2 Oct 2026
    #1 · Preliminary · Stale check · Refresh failed 28 Oct 2026; last good check retained

    471unique models568model configurations134source pages
    Explore all 134 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
    All presets

    Evidence only MultimodalLong contextHuman prefAll capabilitiesThese profiles do not yet have enough comparable evidence for a ranking.

    Adjust weights

    Full ranking

    231 entries
    Compare two configurations

    Value ranking of measured model configurations. Scores are experimental; missing results remain unknown.
    RankModelCapabilityUSD / AA taskEvidence breadth Why this score?
    —ERNIE-4.5-300B-A47B-PT · Non-reasoningBaidu0.8UnknownPrice collection is over 30 days old or dated in the future; refresh before comparing value.Price source ↗Single independent source11 families · 4 areas measured
    —Qwen3-14B · ReasoningQwen0.8UnknownPrice collection is over 30 days old or dated in the future; refresh before comparing value.Single independent source16 families · 5 areas measured
    —Qwen3-VL-8B-Thinking · ReasoningQwen0.8UnknownPrice collection is over 30 days old or dated in the future; refresh before comparing value.Single independent source12 families · 6 areas measured
    —Gemma 4 E4B IT · Non-reasoningGoogle0.8UnknownPrice collection is over 30 days old or dated in the future; refresh before comparing value.Price source ↗Single independent source9 families · 6 areas measured
    —NVIDIA-Nemotron-3-Nano-4B-BF16 · ReasoningNVIDIA0.7UnknownPrice collection is over 30 days old or dated in the future; refresh before comparing value.Single independent source12 families · 5 areas measured
    —Devstral Small 2 · Non-reasoningMistral0.6UnknownPrice collection is over 30 days old or dated in the future; refresh before comparing value.Price source ↗Single independent source17 families · 6 areas measured
    —NVIDIA-Nemotron-Nano-9B-v2 · ReasoningNVIDIA0.6UnknownPrice collection is over 30 days old or dated in the future; refresh before comparing value.Price source ↗Single independent source11 families · 5 areas measured
    —Mistral Medium 3 · Non-reasoningMistral0.6UnknownPrice collection is over 30 days old or dated in the future; refresh before comparing value.Single independent source13 families · 6 areas measured
    —Amazon Nova Premier · Non-reasoningAmazon0.6UnknownPrice collection is over 30 days old or dated in the future; refresh before comparing value.Price source ↗Single independent source13 families · 5 areas measured
    —Qwen3-Coder-30B-A3B-Instruct · Non-reasoningQwen0.6UnknownPrice collection is over 30 days old or dated in the future; refresh before comparing value.Single independent source12 families · 5 areas measured
    —DeepSeek-R1-Distill-Llama-70B · ReasoningDeepSeek0.6UnknownPrice collection is over 30 days old or dated in the future; refresh before comparing value.Single independent source12 families · 5 areas measured
    —Ministral 3 14B · Non-reasoningMistral0.5UnknownPrice collection is over 30 days old or dated in the future; refresh before comparing value.Price source ↗Single independent source17 families · 6 areas measured
    —DeepSeek-R1-0528-Qwen3-8B · ReasoningDeepSeek0.5UnknownPrice collection is over 30 days old or dated in the future; refresh before comparing value.Single independent source12 families · 5 areas measured
    —DeepSeek-R1-Distill-Qwen-32B · ReasoningDeepSeek0.5UnknownPrice collection is over 30 days old or dated in the future; refresh before comparing value.Single independent source8 families · 4 areas measured
    —DeepSeek-R1-Distill-Qwen-14B · ReasoningDeepSeek0.5UnknownPrice collection is over 30 days old or dated in the future; refresh before comparing value.Single independent source8 families · 4 areas measured
    —Qwen3-Omni-30B-A3B-Instruct · Non-reasoningQwen0.5UnknownPrice collection is over 30 days old or dated in the future; refresh before comparing value.Price source ↗Single independent source12 families · 6 areas measured

    More models on this page

    Behind the ranking

    294 entries with published results · 0 ranked · 238 unranked in Value.

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

    Coding rankingReasoning rankingAgentic rankingKnowledge rankingCompare 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: 22 Oct 2026
Mode: value; evidence: documented; configurations: best
Pricing collected: 22 Sept 2026
Cost basis: measured; budget: unlimited USD; input: 1000; output: 1000
https://unifybench.ai/?mode=value&page=8

    Catalog data · Effort measurements · Configuration scoring rules · Methodology

    Catalog dated 22 Oct 2026. Effort evidence collected 18 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 →