New: Agent Index for knowledge work is live.Explore the index ↗

Best LLMs for work automation

AutomationBench

AutomationBench tests agents on realistic business workflows across applications and APIs, covering sales, marketing, operations, support, finance, and HR. It evaluates whether agents can coordinate multi-step work and complete tasks. Zapier’s hosted leaderboard measures strict task completion on a private held-out set.

Updated: Nov 5, 2026Data as of: Nov 2, 2026

Widely usedNot saturated

Official benchmark

AutomationBench results. Each row identifies its benchmark version and evaluation track. Scores from different versions are not directly comparable. Best effort from the latest stored report date per model and version.
Model / creator	Version	Reported score
Gemini 4 CarbonGoogle · Proprietary	1.0.6	53.88%
Gemini 4 ArgonGoogle · Proprietary	1.0.6	51.29%
GLM-5.3-FlashZ.ai · Open source / open weights	1.0.6	48.8%
Claude Sonnet 5.5Anthropic · Proprietary	1.0.6	44.75%
Claude Opus 5.5Anthropic · Proprietary	1.0.6	42.47%
GPT-6 AstraOpenAI · Proprietary	1.0.6	41.4%
Claude Fable 5.1Anthropic · Proprietary	1.0.6	31.4%
Gemini 3.8 FlashGoogle · Proprietary	1.0.6	29.68%
GPT-5.6 SolOpenAI · Proprietary	1.0.6	28.77%
Claude Opus 5Anthropic · Proprietary	1.0.6	26.94%
Kimi K3Moonshot AI · Open source / open weights	1.0.6	22.68%
Muse Spark 1.3Meta · Proprietary	1.0.6	20.7%
Claude Fable 5Anthropic · Proprietary	1.0.6	17.05%
Claude Opus 4.8Anthropic · Proprietary	1.0.6	16.89%
GPT-5.5OpenAI · Proprietary	1.0.6	16.59%
Claude Opus 4.7Anthropic · Proprietary	1.0.6	13.39%
Claude Sonnet 5Anthropic · Proprietary	1.0.6	10.65%
Gemini 3.1 ProGoogle · Proprietary	1.0.6	8.68%

View all 69 published reports

Gemini 4 Carbon53.88%

1.0.6

Gemini 4 Carbon (High); private held-out 657 tasks; API mode; 50-step cap; strict completion

Gemini 4 Carbon52.97%

1.0.6

Gemini 4 Carbon (Medium); private held-out 657 tasks; API mode; 50-step cap; strict completion

Gemini 4 Argon51.29%

1.0.6

Gemini 4 Argon (High); private held-out 657 tasks; API mode; 50-step cap; strict completion

Gemini 4 Argon50.08%

1.0.6

Gemini 4 Argon (Medium); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Sonnet 5.544.75%

1.0.6

Claude Sonnet 5.5 (default fallbacks, Max); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 5.542.47%

1.0.6

Claude Opus 5.5 (default fallbacks, Max); private held-out 657 tasks; API mode; 50-step cap; strict completion

GPT-6 Astra41.4%

1.0.6

GPT-6 Astra (Max); private held-out 657 tasks; API mode; 50-step cap; strict completion

GPT-6 Astra38.96%

1.0.6

GPT-6 Astra (XHigh); private held-out 657 tasks; API mode; 50-step cap; strict completion

GPT-6 Astra37.14%

1.0.6

GPT-6 Astra (High); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Sonnet 5.536.83%

1.0.6

Claude Sonnet 5.5 (default fallbacks, XHigh); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 5.535.77%

1.0.6

Claude Opus 5.5 (default fallbacks, XHigh); private held-out 657 tasks; API mode; 50-step cap; strict completion

GPT-6 Astra34.09%

1.0.6

GPT-6 Astra (Medium); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 5.533.03%

1.0.6

Claude Opus 5.5 (default fallbacks, High); private held-out 657 tasks; API mode; 50-step cap; strict completion

Gemini 4 Carbon31.51%

1.0.6

Gemini 4 Carbon (Low); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Fable 5.131.4%

1.0.6

Claude Fable 5.1 (with Opus 5 Fallback); private held-out 657 tasks; API mode; 50-step cap; strict completion

GPT-6 Astra30.29%

1.0.6

GPT-6 Astra (Low); private held-out 657 tasks; API mode; 50-step cap; strict completion

Gemini 3.8 Flash29.68%

1.0.6

Gemini 3.8 Flash (High); private held-out 657 tasks; API mode; 50-step cap; strict completion

Gemini 3.8 Flash29.68%

1.0.6

Gemini 3.8 Flash (Medium); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 5.529.53%

1.0.6

Claude Opus 5.5 (default fallbacks, Medium); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Sonnet 5.529.38%

1.0.6

Claude Sonnet 5.5 (default fallbacks, High); private held-out 657 tasks; API mode; 50-step cap; strict completion

GPT-5.6 Sol28.77%

1.0.6

GPT-5.6 Sol (Max); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 526.94%

1.0.6

Claude Opus 5 (Max); private held-out 657 tasks; API mode; 50-step cap; strict completion

GPT-5.6 Sol26.33%

1.0.6

GPT-5.6 Sol (XHigh); private held-out 657 tasks; API mode; 50-step cap; strict completion

GPT-6 Astra25.27%

1.0.6

GPT-6 Astra (None); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 525.27%

1.0.6

Claude Opus 5 (XHigh); private held-out 657 tasks; API mode; 50-step cap; strict completion

GPT-5.6 Sol24.81%

1.0.6

GPT-5.6 Sol (High); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 5.524.2%

1.0.6

Claude Opus 5.5 (default fallbacks, Low); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 523.9%

1.0.6

Claude Opus 5 (Medium); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Sonnet 5.523.29%

1.0.6

Claude Sonnet 5.5 (default fallbacks, Medium); private held-out 657 tasks; API mode; 50-step cap; strict completion

1.0.6

Kimi K3; private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Fable 5.122.37%

1.0.6

Claude Fable 5.1 (Max); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Fable 5.121.31%

1.0.6

Claude Fable 5.1 (XHigh); private held-out 657 tasks; API mode; 50-step cap; strict completion

Muse Spark 1.320.7%

1.0.6

Muse Spark 1.3 (Max); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 520.55%

1.0.6

Claude Opus 5 (High); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 520.4%

1.0.6

Claude Opus 5 (Low); private held-out 657 tasks; API mode; 50-step cap; strict completion

GPT-5.6 Sol19.63%

1.0.6

GPT-5.6 Sol (Medium); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Sonnet 5.518.57%

1.0.6

Claude Sonnet 5.5 (default fallbacks, Between tools); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Sonnet 5.518.26%

1.0.6

Claude Sonnet 5.5 (default fallbacks, Low); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Fable 517.05%

1.0.6

Claude Fable 5.0 (Max); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 4.816.89%

1.0.6

Claude Opus 4.8 (XHigh); private held-out 657 tasks; API mode; 50-step cap; strict completion

1.0.6

GPT-5.5 (XHigh); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 4.816.44%

1.0.6

Claude Opus 4.8 (Max); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Fable 516.13%

1.0.6

Claude Fable 5.0 (High); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 4.816.13%

1.0.6

Claude Opus 4.8 (High); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Fable 515.22%

1.0.6

Claude Fable 5.0 (XHigh); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 4.814%

1.0.6

Claude Opus 4.8 (Medium); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Fable 513.7%

1.0.6

Claude Fable 5.0 (Medium); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 4.813.55%

1.0.6

Claude Opus 4.8 (Low); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 4.713.39%

1.0.6

Claude Opus 4.7 (Max); private held-out 657 tasks; API mode; 50-step cap; strict completion

1.0.6

GPT-5.5 (High); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Fable 512.48%

1.0.6

Claude Fable 5.0 (Low); private held-out 657 tasks; API mode; 50-step cap; strict completion

GPT-5.6 Sol11.72%

1.0.6

GPT-5.6 Sol (Low); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Sonnet 510.65%

1.0.6

Claude Sonnet 5 (Max); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Sonnet 59.89%

1.0.6

Claude Sonnet 5 (High); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 4.79.74%

1.0.6

Claude Opus 4.7 (XHigh); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 4.79.13%

1.0.6

Claude Opus 4.7 (High); private held-out 657 tasks; API mode; 50-step cap; strict completion

1.0.6

GPT-5.5 (Medium); private held-out 657 tasks; API mode; 50-step cap; strict completion

Gemini 3.1 Pro8.68%

1.0.6

Gemini 3.1 Pro (preview) (High); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Sonnet 57.61%

1.0.6

Claude Sonnet 5 (XHigh); private held-out 657 tasks; API mode; 50-step cap; strict completion

Gemini 3.1 Pro7%

1.0.6

Gemini 3.1 Pro (preview) (Medium); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 4.76.85%

1.0.6

Claude Opus 4.7 (Medium); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Sonnet 55.78%

1.0.6

Claude Sonnet 5 (Medium); private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Opus 4.75.48%

1.0.6

Claude Opus 4.7 (Low); private held-out 657 tasks; API mode; 50-step cap; strict completion

Gemini 3.1 Pro4.57%

1.0.6

Gemini 3.1 Pro (preview) (Low); private held-out 657 tasks; API mode; 50-step cap; strict completion

GPT-5.6 Sol3.81%

1.0.6

GPT-5.6 Sol; private held-out 657 tasks; API mode; 50-step cap; strict completion

Claude Sonnet 53.65%

1.0.6

Claude Sonnet 5 (Low); private held-out 657 tasks; API mode; 50-step cap; strict completion

1.0.6

GPT-5.5 (Low); private held-out 657 tasks; API mode; 50-step cap; strict completion

1.0.6

GPT-5.5; private held-out 657 tasks; API mode; 50-step cap; strict completion

GLM-5.3-Flash48.8%

1.0.6

harness comparison: provider-reported; settings differ across models · bench_53.png; v1.0.6 with null-type fix PR13; exact task set not established