Benchmark
AutomationBench
Scores are percentages. Higher is better.
- Models and systems measured
- 41
- Scores recorded
- 152
- Sources
- 2
- IQ weight
- 0.00
Score distribution
- 29.8%
- Mean
- 30.3%
- Median
- 13.5%
- Std dev
- 13.5%
- 10th percentile
- 48.8%
- 90th percentile
- 1.2%
- Min
This benchmark has a weight of 0.00 for chat models, measured over 39 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model or system | Score |
|---|---|---|
| 1 | ERNIE 5.5 Flash Preview· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 61.9% |
| 2 | DeepSeek V4.1 Flash 1017· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 60.5% |
| 3 | Mistral Large 4· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 59.9% |
| 4 | DeepSeek V4.1 Flash· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 54.8% |
| 5 | DeepSeek V4.1 FlashDeepSeek V4.1 launch chart · exact public value · setting not further specified· Reported by DeepSeek· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 54.8% |
| 6 | DeepSeek V4.1 Flash官方 scaffold / Pass@1· Reported by DeepSeek· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 54.8% |
| 7 | Gemini 4 Carbon· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 53.9% |
| 8 | Gemini 3.7 Flashv1.0.6 / Z.ai Flash表· Reported by Z.ai· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 52.3% |
| 9 | Claude Fable 5.2· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 51.4% |
| 10 | Gemini 4 Argon· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 51.3% |
| 11 | Gemini 4 ArgonZapier · official launch-page snapshot· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 51.3% |
| 12 | Claude Opus 5DeepSeek V4.1 launch chart · exact public value · setting not further specified· Reported by DeepSeek· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 50.3% |
| 13 | Claude Opus 5max · public v3 · 600 tasks · benchmark automation tools · pass@1· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 50.3% |
| 14 | Muse Spark 1.3max · public v3 · 600 tasks · benchmark automation tools · pass@1· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 49.6% |
| 15 | Muse Spark 1.3· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 49.4% |
| 16 | GLM 5.3DeepSeek V4.1 launch chart · exact public value · setting not further specified· Reported by DeepSeek· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 48.8% |
| 17 | GLM 5.3 Flashsetting not named by the source· Reported by Z.ai· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 48.8% |
| 18 | GLM 5.3 Flash· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 48.8% |
| 19 | GLM 5.3setting not named by the source· Reported by Z.ai· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 48.2% |
| 20 | GLM 5.3· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 48.2% |
| 21 | GPT-5.6 Solmax · public v3 · 600 tasks · benchmark automation tools · pass@1· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 46.7% |
| 22 | Kimi K3setting not named by the source· Reported by Z.ai· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 46.7% |
| 23 | Kimi K3DeepSeek V4.1 launch chart · exact public value · setting not further specified· Reported by DeepSeek· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 46.7% |
| 24 | GPT-5.6 SolDeepSeek V4.1 launch chart · exact public value · setting not further specified· Reported by DeepSeek· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 45.8% |
| 25 | GPT-5.6 Solv1.0.6 / Z.ai设置· Reported by Z.ai· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 45.8% |
| 26 | Claude Sonnet 5.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 44.7% |
| 27 | Claude Sonnet 5.5v1.0.6 private set · max · default fallbacks enabled· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 44.7% |
| 28 | DeepSeek V4 Pro 0813setting not named by the source· Reported by Z.ai· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 43.2% |
| 29 | DeepSeek V4 Pro 0813DeepSeek V4.1 launch chart · exact public value · setting not further specified· Reported by DeepSeek· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 43.2% |
| 30 | Claude Opus 5.5adaptive thinking · max effort · default sampling · 5-trial mean · context ≤1M unless noted · v1.0.6 private set · max · default fallbacks enabled after refusal reruns· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 42.5% |
| 31 | GPT-6 Astra· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 41.4% |
| 32 | GPT-6 AstraAutomationBench v1.0.6 · 47 tools · OpenAI release chart · max reasoning · displayed cost $1.73/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 41.4% |
| 33 | GPT-6 Astramax effort · $1.73 average cost/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 41.4% |
| 34 | GPT-6 AstraOpenAI launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 41.4% |
| 35 | Claude Opus 4.8v1.0.6 / Z.ai设置· Reported by Z.ai· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 41% |
| 36 | Claude Opus 5.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 40% |
| 37 | Claude Opus 5.5Opus 5.5 System Card · adaptive thinking · max effort unless stated · v1.0.6 leaderboard · max effort· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 40% |
| 38 | Beam 1.0 501B A23B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 39.3% |
| 39 | GPT-6 AstraAutomationBench v1.0.6 · 47 tools · OpenAI release chart · xhigh reasoning · displayed cost $1.50/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 39% |
| 40 | GPT-6 Astraxhigh effort · $1.5 average cost/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 39% |
| 41 | Muse Spark 1.2xhigh · public v3 · 600 tasks · benchmark automation tools · pass@1· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 38.2% |
| 42 | DeepSeek V4 Flash 0731DeepSeek V4.1 launch chart · exact public value · setting not further specified· Reported by DeepSeek· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 37.7% |
| 43 | GPT-5.6 Terrav1.0.6 / Z.ai Flash表· Reported by Z.ai· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 37.2% |
| 44 | GPT-6 AstraAutomationBench v1.0.6 · 47 tools · OpenAI release chart · high reasoning · displayed cost $1.44/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 37.1% |
| 45 | GPT-6 Astrahigh effort · $1.44 average cost/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 37.1% |
| 46 | GPT-6.1 SolAutomationBench v1.0.6 · 47 tools · OpenAI release chart · max reasoning · displayed cost $0.30/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 36.1% |
| 47 | GPT-6.1 SolAutomationBench v1.0.6 · 47 tools · OpenAI release chart · xhigh reasoning · displayed cost $0.25/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 35.5% |
| 48 | GPT-6 AstraAutomationBench v1.0.6 · 47 tools · OpenAI release chart · medium reasoning · displayed cost $1.27/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 34.1% |
| 49 | GPT-6 Astramedium effort · $1.27 average cost/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 34.1% |
| 50 | GPT-6 SolAutomationBench v1.0.6 · 47 tools · OpenAI release chart · xhigh reasoning · displayed cost $0.27/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 33.2% |
Showing 1–50 of 145 results.
Not ranked
The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.
| Model or system | Score |
|---|---|
| Claude Fable 5.1 with Opus 5 fallbackAutomationBench v1.0.6 · 47 tools · OpenAI release chart · max reasoning · displayed cost $2.45/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 31.4% |
| Claude Fable 5.1 with Opus 5 fallbackmax effort · $2.45 average cost/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 31.4% |
| Claude Opus 5.5 with fallbacks (OpenAI chart label)AutomationBench v1.0.6 · 47 tools · OpenAI release chart · high reasoning · displayed cost $0.71/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 33% |
| Claude Opus 5.5 with fallbacks (OpenAI chart label)AutomationBench v1.0.6 · 47 tools · OpenAI release chart · low reasoning · displayed cost $0.51/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 24.2% |
| Claude Opus 5.5 with fallbacks (OpenAI chart label)AutomationBench v1.0.6 · 47 tools · OpenAI release chart · max reasoning · displayed cost $1.44/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 42.5% |
| Claude Opus 5.5 with fallbacks (OpenAI chart label)AutomationBench v1.0.6 · 47 tools · OpenAI release chart · medium reasoning · displayed cost $0.65/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 29.5% |
| Claude Opus 5.5 with fallbacks (OpenAI chart label)AutomationBench v1.0.6 · 47 tools · OpenAI release chart · xhigh reasoning · displayed cost $0.89/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 35.8% |
Data credits: Data by LLM Stats · BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.