CtrlAltDebrief

AI model benchmarks

25 benchmarks across reasoning, maths, coding and agentic tasks. Each one ranks every model that carries a score and puts its API cost alongside, where it has one.

Benchmarks

25

leaderboards

Categories

5

reasoning · maths · coding · agents

Models ranked

411

on the widest board

Source

AA + HF

verified daily

Headline indices

3 benchmarks
Intelligence IndexComposite
411 models ranked
  1. 1Claude Fable 559.9
  2. 2GPT-5.6 Sol58.9
  3. 3Claude Opus 4.855.7
View full leaderboard →
Coding IndexComposite
136 models ranked
  1. 1GPT-5.6 Sol78.3
  2. 2GPT-5.6 Terra76.7
  3. 3Claude Fable 576.5
View full leaderboard →
Agentic IndexComposite
111 models ranked
  1. 1GPT-5.6 Sol54.0
  2. 2Claude Fable 552.8
  3. 3GPT-5.6 Terra47.4
View full leaderboard →

Reasoning & knowledge

7 benchmarks
GPQA Diamond
398 models ranked
  1. 1Gemini 3.1 Pro Preview94%
  2. 2GPT-5.6 Sol94%
  3. 3GPT-5.594%
View full leaderboard →
Humanity's Last Exam
388 models ranked
  1. 1Claude Fable 553%
  2. 2GPT-5.6 Sol47%
  3. 3Claude Opus 4.846%
View full leaderboard →
CritPt
307 models ranked
  1. 1GPT-5.6 Sol32%
  2. 2GPT-5.5 Pro31%
  3. 3GPT-5.6 Terra30%
View full leaderboard →
Omniscience
303 models ranked
  1. 1Claude Fable 540.1
  2. 2Gemini 3.1 Pro Preview32.9
  3. 3Claude Opus 4.827.4
View full leaderboard →
MMLU-Pro
266 models ranked
  1. 1Gemini 3 Pro90%
  2. 2Claude Opus 4.589%
  3. 3Gemini 3 Flash89%
View full leaderboard →
MMMU-Pro
137 models ranked
  1. 1Gemini 3.5 Flash84%
  2. 2GPT-5.6 Sol83%
  3. 3Gemini 3.1 Pro Preview82%
View full leaderboard →
ARC-AGI-2
61 models ranked
  1. 1GPT-5.6 Sol93%
  2. 2GPT-5.585%
  3. 3GPT-5.5 Pro85%
View full leaderboard →

Math

1 benchmarks
AIME 2026
14 models ranked
  1. 1Step 3.5 Flash 260397%
  2. 2Kimi K2.696%
  3. 3Kimi K2.596%
View full leaderboard →

Coding & agents

12 benchmarks
SciCode
386 models ranked
  1. 1Claude Fable 560%
  2. 2Gemini 3.1 Pro Preview59%
  3. 3Muse Spark 1.158%
View full leaderboard →
τ²-bench
308 models ranked
  1. 1GLM 5.299%
  2. 2JT-35B-Flash99%
  3. 3GLM 4.7 Flash99%
View full leaderboard →
TerminalBench Hard
302 models ranked
  1. 1GPT-5.6 Sol66%
  2. 2Claude Fable 563%
  3. 3GPT-5.6 Terra63%
View full leaderboard →
LiveCodeBench
265 models ranked
  1. 1Gemini 3 Pro92%
  2. 2Gemini 3 Flash91%
  3. 3DeepSeek V3 2 Speciale90%
View full leaderboard →
τ-bench Banking
117 models ranked
  1. 1GPT-5.6 Sol33%
  2. 2Grok 4.533%
  3. 3GPT-5.6 Terra32%
View full leaderboard →
TerminalBench v2.1
116 models ranked
  1. 1GPT-5.6 Sol90%
  2. 2GPT-5.6 Terra88%
  3. 3Claude Opus 4.885%
View full leaderboard →
GDPval
115 models ranked
  1. 1Claude Fable 51759.6
  2. 2GPT-5.6 Sol1747.8
  3. 3Claude Sonnet 51606.7
View full leaderboard →
IT-Bench SRE
31 models ranked
  1. 1GPT-5.6 Sol56%
  2. 2GPT-5.6 Terra51%
  3. 3Claude Opus 4.747%
View full leaderboard →
SWE-bench Verified
31 models ranked
  1. 1DeepSeek V4 Pro81%
  2. 2MiniMax M381%
  3. 3Kimi K2.680%
View full leaderboard →
APEX Agents
26 models ranked
  1. 1Gemini 3.5 Flash47%
  2. 2GPT-5.538%
  3. 3GLM 5.234%
View full leaderboard →
AA-Briefcase
24 models ranked
  1. 1Claude Fable 554%
  2. 2Claude Sonnet 545%
  3. 3Claude Opus 4.843%
View full leaderboard →
SWE-bench Pro
22 models ranked
  1. 1GLM 5.262%
  2. 2MiniMax M359%
  3. 3Kimi K2.659%
View full leaderboard →

Instruction & long-context

2 benchmarks
Long-Context ReasoningComposite
322 models ranked
  1. 1GPT-5.2-Codex76%
  2. 2GPT-576%
  3. 3GPT-5.175%
View full leaderboard →
IFBench
314 models ranked
  1. 1Grok 4.383%
  2. 2Grok 4.20 030983%
  3. 3MiniMax M383%
View full leaderboard →
Benchmark scores via ARC Prize, Artificial Analysis and Hugging Face leaderboards · pricing verified against LiteLLM + OpenRouter · refreshed daily

Every weekday

AI moves fast. Here's your debrief.

News, analysis, tools, and more.

For people who build with AI