2026-09-10MODELDeepSeek-V4.1-Flash released with extreme agent memory efficiency and evaluated across VALS • 2026-09-03MODELGPT-6 Astra released by OpenAI - 71.7% on EMB, 63.3% on Tax Agent Bench • 2026-09-02MODELGoogle Gemini 3.8 Flash takes #1 rank on Finance Agent (v2) with 61.4% and 72.2% on EMB • 2026-09-02MODELMeta releases Muse Spark 1.3 & 1.3 Max - 59.0%-60.0% on Finance Agent (v2) • 2026-09-01MODELClaude Fable 5.1 leads Tax Agent Bench (77.6%) and EMB (76.7%) with 75% cheaper cache reads • 2026-09-11BENCHTax Agent Bench and FAB v2 benchmarks updated with latest frontier model rankings • 2026-09-15UPDATEPlatform updated: 43 models tracked across 41 independent benchmarks • 2026-09-10MODELDeepSeek-V4.1-Flash released with extreme agent memory efficiency and evaluated across VALS • 2026-09-03MODELGPT-6 Astra released by OpenAI - 71.7% on EMB, 63.3% on Tax Agent Bench • 2026-09-02MODELGoogle Gemini 3.8 Flash takes #1 rank on Finance Agent (v2) with 61.4% and 72.2% on EMB • 2026-09-02MODELMeta releases Muse Spark 1.3 & 1.3 Max - 59.0%-60.0% on Finance Agent (v2) • 2026-09-01MODELClaude Fable 5.1 leads Tax Agent Bench (77.6%) and EMB (76.7%) with 75% cheaper cache reads • 2026-09-11BENCHTax Agent Bench and FAB v2 benchmarks updated with latest frontier model rankings • 2026-09-15UPDATEPlatform updated: 43 models tracked across 41 independent benchmarks •
Home/Benchmarks/MMLU Pro
Tier 2numericalverification

MMLU Pro

Robust multi-task language understanding reasoning benchmark with reasoning-focused questions.

12,000 tasks
21 models evaluated
Updated 2026
arXiv:2406.01574
View Source →Submit your scoreMethodology
91.6%
Top Score
Top model
Claude Opus 5
Model Rankings
financebenchmark.ai · MMLU Pro
Anthropic
Anthropic
OpenAI
DeepSeek
xAI
Moonshot AI
Anthropic
DeepSeek
Google
Google
Zhipu AI
OpenAI
OpenAI
xAI
Thinking Machines
Thinking Machines
Ant Group
Meta
Meta
NVIDIA
Alibaba
88.0%
🥇AnthropicAnthropicClaude Opus 591.6%
🥈AnthropicAnthropicClaude Sonnet 587.5%
🥉OpenAIOpenAIGPT-5.6 Sol89.1%
4DeepSeekDeepSeekDeepSeek V4 FlashOPEN86.2%
5xAIxAIGrok 4.689.4%
6Moonshot AIMoonshot AIKimi K3OPEN88.0%
7AnthropicAnthropicClaude Fable 591.5%
8DeepSeekDeepSeekDeepSeek V4 Pro (0813)OPEN87.0%
9GoogleGoogleGemini 3.5 Flash Lite85.8%
10GoogleGoogleGemini 3.6 Flash88.5%
11Zhipu AIZhipu AIGLM 5.2OPEN86.7%
12OpenAIOpenAIGPT-5.6 Luna86.0%
13OpenAIOpenAIGPT-5.6 Terra86.7%
14xAIxAIGrok 4.589.2%
15Thinking MachinesThinking MachinesInklingOPEN86.3%
16Thinking MachinesThinking MachinesInkling SmallOPEN85.6%
17Ant GroupAnt GroupLing 3.0 FlashOPEN82.0%
18MetaMetaMuse Spark 1.188.7%
19MetaMetaMuse Spark 1.288.3%
20NVIDIANVIDIANemotron 3 UltraOPEN85.8%
21AlibabaAlibabaQwen 3.8 MaxOPEN88.6%
Full Results Table
#ModelOverallSourceDate
🥇
AnthropicClaude Opus 5
Anthropic
91.6arXiv:2406.015742026-07
🥈
AnthropicClaude Fable 5
Anthropic
91.5arXiv:2406.015742026-06
🥉
xAIGrok 4.6
xAI
89.4arXiv:2406.015742026-08
4
xAIGrok 4.5
xAI
89.2arXiv:2406.015742026-07
5
OpenAIGPT-5.6 Sol
OpenAI
89.1arXiv:2406.015742026-07
6
MetaMuse Spark 1.1
Meta
88.7arXiv:2406.015742026-07
7
AlibabaQwen 3.8 MaxOPEN
Alibaba
88.6arXiv:2406.015742026-08
8
GoogleGemini 3.6 Flash
Google
88.5arXiv:2406.015742026-07
9
MetaMuse Spark 1.2
Meta
88.3arXiv:2406.015742026-08
10
Moonshot AIKimi K3OPEN
Moonshot AI
88.0arXiv:2406.015742026-07
11
AnthropicClaude Sonnet 5
Anthropic
87.5arXiv:2406.015742026-06
12
DeepSeekDeepSeek V4 Pro (0813)OPEN
DeepSeek
87.0arXiv:2406.015742026-08
13
Zhipu AIGLM 5.2OPEN
Zhipu AI
86.7arXiv:2406.015742026-06
14
OpenAIGPT-5.6 Terra
OpenAI
86.7arXiv:2406.015742026-07
15
Thinking MachinesInklingOPEN
Thinking Machines
86.3arXiv:2406.015742026-07
16
DeepSeekDeepSeek V4 FlashOPEN
DeepSeek
86.2arXiv:2406.015742026-07
17
OpenAIGPT-5.6 Luna
OpenAI
86.0arXiv:2406.015742026-07
18
GoogleGemini 3.5 Flash Lite
Google
85.8arXiv:2406.015742026-07
19
NVIDIANemotron 3 UltraOPEN
NVIDIA
85.8arXiv:2406.015742026-06
20
Thinking MachinesInkling SmallOPEN
Thinking Machines
85.6arXiv:2406.015742026-07
21
Ant GroupLing 3.0 FlashOPEN
Ant Group
82.0arXiv:2406.015742026-07
All evaluated models
Anthropic
Claude Opus 5
91.6%
Anthropic
Claude Sonnet 5
87.5%
OpenAI
GPT-5.6 Sol
89.1%
DeepSeek
DeepSeek V4 Flash
86.2%
xAI
Grok 4.6
89.4%
Moonshot AI
Kimi K3
88.0%
Anthropic
Claude Fable 5
91.5%
DeepSeek
DeepSeek V4 Pro (0813)
87.0%
Google
Gemini 3.5 Flash Lite
85.8%
Google
Gemini 3.6 Flash
88.5%
Zhipu AI
GLM 5.2
86.7%
OpenAI
GPT-5.6 Luna
86.0%
OpenAI
GPT-5.6 Terra
86.7%
xAI
Grok 4.5
89.2%
Thinking Machines
Inkling
86.3%
Thinking Machines
Inkling Small
85.6%
Ant Group
Ling 3.0 Flash
82.0%
Meta
Muse Spark 1.1
88.7%
Meta
Muse Spark 1.2
88.3%
NVIDIA
Nemotron 3 Ultra
85.8%
Alibaba
Qwen 3.8 Max
88.6%