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/ProofBench v1.1
Tier 3numericalverification

ProofBench v1.1

Formal mathematical theorem proving and step-by-step rigorous logical deduction.

Variable tasks
13 models evaluated
Updated 2026
VALS (2026)
View Source →Submit your scoreMethodology
99.0%
Top Score
Top model
Claude Opus 5
Model Rankings
financebenchmark.ai · ProofBench
Anthropic
Anthropic
OpenAI
DeepSeek
Moonshot AI
Anthropic
OpenAI
OpenAI
xAI
Thinking Machines
Thinking Machines
Meta
Alibaba
95.0%
🥇AnthropicAnthropicClaude Opus 599.0%
🥈AnthropicAnthropicClaude Sonnet 577.0%
🥉OpenAIOpenAIGPT-5.6 Sol83.0%
4DeepSeekDeepSeekDeepSeek V4 FlashOPEN56.0%
5Moonshot AIMoonshot AIKimi K3OPEN87.0%
6AnthropicAnthropicClaude Fable 595.0%
7OpenAIOpenAIGPT-5.6 Luna60.0%
8OpenAIOpenAIGPT-5.6 Terra71.0%
9xAIxAIGrok 4.531.0%
10Thinking MachinesThinking MachinesInklingOPEN0.0%
11Thinking MachinesThinking MachinesInkling SmallOPEN6.0%
12MetaMetaMuse Spark 1.243.0%
13AlibabaAlibabaQwen 3.8 MaxOPEN58.0%
Full Results Table
#ModelOverallSourceDate
🥇
AnthropicClaude Opus 5
Anthropic
99.0VALS (2026)2026-07
🥈
AnthropicClaude Fable 5
Anthropic
95.0VALS (2026)2026-06
🥉
Moonshot AIKimi K3OPEN
Moonshot AI
87.0VALS (2026)2026-07
4
OpenAIGPT-5.6 Sol
OpenAI
83.0VALS (2026)2026-07
5
AnthropicClaude Sonnet 5
Anthropic
77.0VALS (2026)2026-06
6
OpenAIGPT-5.6 Terra
OpenAI
71.0VALS (2026)2026-07
7
OpenAIGPT-5.6 Luna
OpenAI
60.0VALS (2026)2026-07
8
AlibabaQwen 3.8 MaxOPEN
Alibaba
58.0VALS (2026)2026-08
9
DeepSeekDeepSeek V4 FlashOPEN
DeepSeek
56.0VALS (2026)2026-07
10
MetaMuse Spark 1.2
Meta
43.0VALS (2026)2026-08
11
xAIGrok 4.5
xAI
31.0VALS (2026)2026-07
12
Thinking MachinesInkling SmallOPEN
Thinking Machines
6.0VALS (2026)2026-07
13
Thinking MachinesInklingOPEN
Thinking Machines
0.0VALS (2026)2026-07
All evaluated models
Anthropic
Claude Opus 5
99.0%
Anthropic
Claude Sonnet 5
77.0%
OpenAI
GPT-5.6 Sol
83.0%
DeepSeek
DeepSeek V4 Flash
56.0%
Moonshot AI
Kimi K3
87.0%
Anthropic
Claude Fable 5
95.0%
OpenAI
GPT-5.6 Luna
60.0%
OpenAI
GPT-5.6 Terra
71.0%
xAI
Grok 4.5
31.0%
Thinking Machines
Inkling
0.0%
Thinking Machines
Inkling Small
6.0%
Meta
Muse Spark 1.2
43.0%
Alibaba
Qwen 3.8 Max
58.0%