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/Finance Agent Benchmark
Tier 1agentTasks

Finance Agent Benchmark

Core financial analyst tasks and equity research evaluated using the Finance Agent (FAB v2) harness.

Variable tasks
40 models evaluated
Updated 2026
arXiv:2508.00828
View Source →Submit your scoreMethodology
61.4%
Top Score
Top model
Gemini 3.8 Flash
Model Rankings
financebenchmark.ai · Finance Agent
Anthropic
OpenAI
Google
DeepSeek
xAI
Moonshot AI
Zhipu AI
MiniMax
Anthropic
Anthropic
OpenAI
Google
DeepSeek
xAI
Moonshot AI
MiniMax
Anthropic
DeepSeek
Google
Google
Zhipu AI
OpenAI
OpenAI
xAI
Thinking Machines
Thinking Machines
Ant Group
Meta
Meta
NVIDIA
Alibaba
Alibaba
OpenAI
Anthropic
Google
Meta
Meta
DeepSeek
Zhipu AI
Zhipu AI
44.9%
🥇AnthropicAnthropicClaude Opus 4.751.5%
🥈OpenAIOpenAIGPT-5.551.8%
🥉GoogleGoogleGemini 3.1 Pro43.0%
4DeepSeekDeepSeekDeepSeek V4 ProOPEN44.1%
5xAIxAIGrok 4.337.7%
6Moonshot AIMoonshot AIKimi K2.6OPEN44.9%
7Zhipu AIZhipu AIGLM-5.1OPEN44.8%
8MiniMaxMiniMaxMiniMax M2.7OPEN27.9%
9AnthropicAnthropicClaude Opus 558.6%
10AnthropicAnthropicClaude Sonnet 553.9%
11OpenAIOpenAIGPT-5.6 Sol53.8%
12GoogleGoogleGemini 3.7 Flash59.0%
13DeepSeekDeepSeekDeepSeek V4 FlashOPEN49.5%
14xAIxAIGrok 4.653.7%
15Moonshot AIMoonshot AIKimi K3OPEN54.4%
16MiniMaxMiniMaxMiniMax M3OPEN48.3%
17AnthropicAnthropicClaude Fable 556.3%
18DeepSeekDeepSeekDeepSeek V4 Pro (0813)OPEN50.4%
19GoogleGoogleGemini 3.5 Flash Lite47.4%
20GoogleGoogleGemini 3.6 Flash56.3%
21Zhipu AIZhipu AIGLM 5.2OPEN49.7%
22OpenAIOpenAIGPT-5.6 Luna55.0%
23OpenAIOpenAIGPT-5.6 Terra54.4%
24xAIxAIGrok 4.548.4%
25Thinking MachinesThinking MachinesInklingOPEN46.6%
26Thinking MachinesThinking MachinesInkling SmallOPEN41.3%
27Ant GroupAnt GroupLing 3.0 FlashOPEN30.3%
28MetaMetaMuse Spark 1.157.2%
29MetaMetaMuse Spark 1.260.6%
30NVIDIANVIDIANemotron 3 UltraOPEN37.7%
31AlibabaAlibabaQwen 3.7 Plus38.2%
32AlibabaAlibabaQwen 3.8 MaxOPEN50.6%
33OpenAIOpenAIGPT-6 Astra53.5%
34AnthropicAnthropicClaude Fable 5.158.9%
35GoogleGoogleGemini 3.8 Flash61.4%
36MetaMetaMuse Spark 1.358.9%
37MetaMetaMuse Spark 1.3 Max60.0%
38DeepSeekDeepSeekDeepSeek V4.1 FlashOPEN53.5%
39Zhipu AIZhipu AIGLM 5.3OPEN55.8%
40Zhipu AIZhipu AIGLM 5.3 FlashOPEN57.9%
Full Results Table
#ModelOverallSourceDate
🥇
GoogleGemini 3.8 Flash
Google
61.4arXiv:2508.008282026-09
🥈
MetaMuse Spark 1.2
Meta
60.6arXiv:2508.008282026-09
🥉
MetaMuse Spark 1.3 Max
Meta
60.0arXiv:2508.008282026-09
4
GoogleGemini 3.7 Flash
Google
59.0arXiv:2508.008282026-09
5
MetaMuse Spark 1.3
Meta
58.9arXiv:2508.008282026-09
6
AnthropicClaude Fable 5.1
Anthropic
58.9arXiv:2508.008282026-09
7
AnthropicClaude Opus 5
Anthropic
58.6arXiv:2508.008282026-09
8
Zhipu AIGLM 5.3 FlashOPEN
Zhipu AI
57.9arXiv:2508.008282026-09
9
MetaMuse Spark 1.1
Meta
57.2arXiv:2508.008282026-09
10
AnthropicClaude Fable 5
Anthropic
56.3arXiv:2508.008282026-09
11
GoogleGemini 3.6 Flash
Google
56.3arXiv:2508.008282026-09
12
Zhipu AIGLM 5.3OPEN
Zhipu AI
55.8arXiv:2508.008282026-09
13
OpenAIGPT-5.6 Luna
OpenAI
55.0arXiv:2508.008282026-09
14
OpenAIGPT-5.6 Terra
OpenAI
54.4arXiv:2508.008282026-09
15
Moonshot AIKimi K3OPEN
Moonshot AI
54.4arXiv:2508.008282026-09
16
AnthropicClaude Sonnet 5
Anthropic
53.9arXiv:2508.008282026-09
17
OpenAIGPT-5.6 Sol
OpenAI
53.8arXiv:2508.008282026-09
18
xAIGrok 4.6
xAI
53.7arXiv:2508.008282026-09
19
OpenAIGPT-6 Astra
OpenAI
53.5arXiv:2508.008282026-09
20
DeepSeekDeepSeek V4.1 FlashOPEN
DeepSeek
53.5arXiv:2508.008282026-09
21
OpenAIGPT-5.5
OpenAI
51.8arXiv:2508.008282026-09
22
AnthropicClaude Opus 4.7
Anthropic
51.5arXiv:2508.008282026-09
23
AlibabaQwen 3.8 MaxOPEN
Alibaba
50.6arXiv:2508.008282026-09
24
DeepSeekDeepSeek V4 Pro (0813)OPEN
DeepSeek
50.4arXiv:2508.008282026-09
25
Zhipu AIGLM 5.2OPEN
Zhipu AI
49.7arXiv:2508.008282026-09
26
DeepSeekDeepSeek V4 FlashOPEN
DeepSeek
49.5arXiv:2508.008282026-09
27
xAIGrok 4.5
xAI
48.4arXiv:2508.008282026-09
28
MiniMaxMiniMax M3OPEN
MiniMax
48.3arXiv:2508.008282026-09
29
GoogleGemini 3.5 Flash Lite
Google
47.4arXiv:2508.008282026-09
30
Thinking MachinesInklingOPEN
Thinking Machines
46.6arXiv:2508.008282026-09
31
Moonshot AIKimi K2.6OPEN
Moonshot AI
44.9arXiv:2508.008282026-09
32
Zhipu AIGLM-5.1OPEN
Zhipu AI
44.8arXiv:2508.008282026-09
33
DeepSeekDeepSeek V4 ProOPEN
DeepSeek
44.1arXiv:2508.008282026-09
34
GoogleGemini 3.1 Pro
Google
43.0arXiv:2508.008282026-09
35
Thinking MachinesInkling SmallOPEN
Thinking Machines
41.3arXiv:2508.008282026-09
36
AlibabaQwen 3.7 Plus
Alibaba
38.2arXiv:2508.008282026-09
37
xAIGrok 4.3
xAI
37.7arXiv:2508.008282026-09
38
NVIDIANemotron 3 UltraOPEN
NVIDIA
37.7arXiv:2508.008282026-09
39
Ant GroupLing 3.0 FlashOPEN
Ant Group
30.3arXiv:2508.008282026-09
40
MiniMaxMiniMax M2.7OPEN
MiniMax
27.9arXiv:2508.008282026-09
All evaluated models
Anthropic
Claude Opus 4.7
51.5%
OpenAI
GPT-5.5
51.8%
Google
Gemini 3.1 Pro
43.0%
DeepSeek
DeepSeek V4 Pro
44.1%
xAI
Grok 4.3
37.7%
Moonshot AI
Kimi K2.6
44.9%
Zhipu AI
GLM-5.1
44.8%
MiniMax
MiniMax M2.7
27.9%
Anthropic
Claude Opus 5
58.6%
Anthropic
Claude Sonnet 5
53.9%
OpenAI
GPT-5.6 Sol
53.8%
Google
Gemini 3.7 Flash
59.0%
DeepSeek
DeepSeek V4 Flash
49.5%
xAI
Grok 4.6
53.7%
Moonshot AI
Kimi K3
54.4%
MiniMax
MiniMax M3
48.3%
Anthropic
Claude Fable 5
56.3%
DeepSeek
DeepSeek V4 Pro (0813)
50.4%
Google
Gemini 3.5 Flash Lite
47.4%
Google
Gemini 3.6 Flash
56.3%
Zhipu AI
GLM 5.2
49.7%
OpenAI
GPT-5.6 Luna
55.0%
OpenAI
GPT-5.6 Terra
54.4%
xAI
Grok 4.5
48.4%
Thinking Machines
Inkling
46.6%
Thinking Machines
Inkling Small
41.3%
Ant Group
Ling 3.0 Flash
30.3%
Meta
Muse Spark 1.1
57.2%
Meta
Muse Spark 1.2
60.6%
NVIDIA
Nemotron 3 Ultra
37.7%
Alibaba
Qwen 3.7 Plus
38.2%
Alibaba
Qwen 3.8 Max
50.6%
OpenAI
GPT-6 Astra
53.5%
Anthropic
Claude Fable 5.1
58.9%
Google
Gemini 3.8 Flash
61.4%
Meta
Muse Spark 1.3
58.9%
Meta
Muse Spark 1.3 Max
60.0%
DeepSeek
DeepSeek V4.1 Flash
53.5%
Zhipu AI
GLM 5.3
55.8%
Zhipu AI
GLM 5.3 Flash
57.9%