BenchmarkActivecategory · generalfamily · livebench · variant global

LiveBench

livebench.ai

contamination-limited, monthly refreshed questions across 7 categories — overall mean of the category averages

quality57

Updated 28 min ago · first seen 11 Sept 2026

Metric: global_average · % ↑
Current results: 57
Models: 57
Current leader: Claude Fable 5.1 83.4%

Score history · Gemini 3.5 Flash-Lite High 8 rows

Not enough history to chart — 8 observations, all dated 25 Jun 2026. Rows under different configurations count separately; the list below shows each one.

67.24%release=2026-06-25 · subtasks=["paraphrase","simplify","story_generation","summarize"] · livebench_model_id=gemini-3.5-flash-lite-high25 Jun 2026
71.82%release=2026-06-25 · subtasks=["connections","plot_unscrambling","typos"] · livebench_model_id=gemini-3.5-flash-lite-high25 Jun 2026
53.25%release=2026-06-25 · subtasks=["consecutive_events","tablejoin","tablereformat"] · livebench_model_id=gemini-3.5-flash-lite-high25 Jun 2026
73.74%release=2026-06-25 · subtasks=["AMPS_Hard","integrals_with_game","math_comp","olympiad"] · livebench_model_id=gemini-3.5-flash-lite-high25 Jun 2026
45.25%release=2026-06-25 · subtasks=["javascript","typescript","python"] · livebench_model_id=gemini-3.5-flash-lite-high25 Jun 2026
76.07%release=2026-06-25 · subtasks=["code_generation","code_completion"] · livebench_model_id=gemini-3.5-flash-lite-high25 Jun 2026
60.19%release=2026-06-25 · subtasks=["theory_of_mind","zebra_puzzle","spatial","logic_with_navigation"] · livebench_model_id=gemini-3.5-flash-lite-high25 Jun 2026
63.94%release=2026-06-25 · aggregation=mean of category averages; category = mean of its subtasks · livebench_model_id=gemini-3.5-flash-lite-high25 Jun 2026

Back to the leaderboard

Frontier over time · global_average

9 leader changes recorded, all dated 25 Jun 2026 — the frontier line needs at least two distinct dates. The corpus is young: every result was first observed on the same day, so leader changes will separate in time as sources are re-crawled.

83.4%Claude Fable 5.1 Anthropic Official board25 Jun 2026
83.0%Claude Fable 5 Anthropic Official board25 Jun 2026
81.1%gpt-5.6-sol OpenAI Official board25 Jun 2026
80.2%gpt-5.5 OpenAI Official board25 Jun 2026
78.0%gpt-5.4 OpenAI Official board25 Jun 2026
77.0%Gemini 3.1 Pro Preview Google Official board25 Jun 2026
76.5%Claude Opus 4.7 Anthropic Official board25 Jun 2026
74.5%Claude Opus 4.6 Anthropic Official board25 Jun 2026

Includes closed rows (history). A point is emitted whenever a result beats every earlier result of the same group, ordered by evaluated_at when the source publishes it, else observed_at.

Leaderboard 57 models

Select models with +, then Compare.

No result in this group with these filters

Relax the trust / organization filters or pick another comparability group.

Definition

Task: Source:AI Atlas curated registry (YAML, versioned in git, every entry carries its source URL)T2observed 28 min agomedium
Metric: Source:AI Atlas curated registry (YAML, versioned in git, every entry carries its source URL)T2observed 28 min agomedium
Metric range: Source:AI Atlas curated registry (YAML, versioned in git, every entry carries its source URL)T2observed 28 min agomedium
Variant: Source:AI Atlas curated registry (YAML, versioned in git, every entry carries its source URL)T2observed 28 min agomedium
Harness: Source:AI Atlas curated registry (YAML, versioned in git, every entry carries its source URL)T2observed 28 min agomedium
Comparability note: Source:AI Atlas curated registry (YAML, versioned in git, every entry carries its source URL)T2observed 28 min agomedium
Paper: Source:AI Atlas curated registry (YAML, versioned in git, every entry carries its source URL)T2observed 10 h agomedium
Website: livebench.ai
Source:AI Atlas curated registry (YAML, versioned in git, every entry carries its source URL)T2observed 10 h agomedium

Active group

global_average

57 models · 57 current rows

Comparability

Comparable same task and conditions · Partially comparable same task, conditions differ (effort, temperature, judge) · Not comparable different variant or metric

Also known as

LiveBench, LiveBench Global, LiveBench global average, LiveBench overall

slug livebench-2 · bench_01M293SPERF0D8QG7TSCGE8GGF

Rows are grouped by comparability group = canonical metric × config_key (hash of the task-defining configuration keys: variant, evaluator, harness, shots, pass regime…). A leaderboard shows one row per canonical model: its best current row inside the group. Reasoning effort, temperature or judge differences keep rows in the same group but mark them partially comparable. delta_rank compares with the ranking built from the closed (previous) rows of the same group.