Benchmark
JobBench
Agentic benchmarks measure how well models carry out tasks that take several steps. Scores are percentages. Higher is better.
- Models and systems measured
- 27
- Scores recorded
- 40
- Sources
- 2
- IQ weight
- 1.75
Score distribution
- 53.4%
- Mean
- 54.7%
- Median
- 13.1%
- Std dev
- 33.4%
- 10th percentile
- 65.7%
- 90th percentile
- 15.9%
- Min
This benchmark has a weight of 1.75 for chat models, measured over 26 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 | Muse Wonder Full· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 75.3% |
| 2 | Muse Wonder· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 70.5% |
| 3 | Muse Wonder 1.0· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 70.5% |
| 4 | Claude Opus 5Atria Dawn Preview official card / report Table 1 · published comparison run · model snapshots and reasoning tiers not specified unless stated · official OpenCode harness · main split / 65 tasks · 7,200-second limit · two attempts · Grok-4.3 judge · mean task score· Reported by Shanghai AI Laboratory· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 68% |
| 5 | Claude Opus 5max · 65 tasks · official evaluation code · OpenCode · file-aware rubric grader · mean rubric score· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 65.7% |
| 6 | LongCat 2.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 65.4% |
| 7 | Muse Spark 1.3· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 64.9% |
| 8 | Muse Spark 1.3max · 65 tasks · official evaluation code · OpenCode · file-aware rubric grader · mean rubric score· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 64.9% |
| 9 | MiMo V2.6 Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 62% |
| 10 | Muse Spark Open· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 61.8% |
| 11 | Hy4 preview· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 61.7% |
| 12 | Hy4 previewofficial model-card appendix· Reported by Tencent Hy· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 61.7% |
| 13 | Muse Spark 1.2xhigh · 65 tasks · official evaluation code · OpenCode · file-aware rubric grader · mean rubric score· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 61.6% |
| 14 | MiMo V2.6 Flash· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 61.2% |
| 15 | GLM 5.3Atria Dawn Preview official card / report Table 1 · published comparison run · model snapshots and reasoning tiers not specified unless stated · official OpenCode harness · main split / 65 tasks · 7,200-second limit · two attempts · Grok-4.3 judge · mean task score· Reported by Shanghai AI Laboratory· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 58.2% |
| 16 | Claude Fable 5Qwen3.8 官方模型卡对照表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 57.4% |
| 17 | Qwen 3.8 Flash Next· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 55.7% |
| 18 | Qwen 3.8 Flash NextQwen3.8-Flash-Next release/model card · published setting· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 55.7% |
| 19 | Qwen3.8 Flash· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 55.7% |
| 20 | Muse Spark 1.1· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 54.7% |
| 21 | Muse Spark 1.1Meta Model API · xhigh · OpenCode · Grok 4.3 judge · official JobBench results· Reported by Meta, Meta Superintelligence Labs· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 54.7% |
| 22 | Kimi K3Atria Dawn Preview official card / report Table 1 · published comparison run · model snapshots and reasoning tiers not specified unless stated · official OpenCode harness · main split / 65 tasks · 7,200-second limit · two attempts · Grok-4.3 judge · mean task score· Reported by Shanghai AI Laboratory· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 54.3% |
| 23 | Kimi K3max effort · temp 1.0 · top_p 1.0· Reported by Moonshot AI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 54.3% |
| 24 | DeepSeek V4 Pro 0813Atria Dawn Preview official card / report Table 1 · published comparison run · model snapshots and reasoning tiers not specified unless stated · official OpenCode harness · main split / 65 tasks · 7,200-second limit · two attempts · Grok-4.3 judge · mean task score· Reported by Shanghai AI Laboratory· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 54.1% |
| 25 | Qwen3.8 Max· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 53.4% |
| 26 | Qwen3.8 MaxQwen3.8 官方模型卡对照表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 53.4% |
| 27 | Kimi K3· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 52.9% |
| 28 | Qwen3.8 MaxAtria Dawn Preview official card / report Table 1 · published comparison run · model snapshots and reasoning tiers not specified unless stated · official OpenCode harness · main split / 65 tasks · 7,200-second limit · two attempts · Grok-4.3 judge · mean task score· Reported by Shanghai AI Laboratory· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 52.7% |
| 29 | Inkling 2 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 49.6% |
| 30 | Claude Opus 4.8max · OpenCode · Grok 4.3 judge · official JobBench results· Reported by Meta, Meta Superintelligence Labs· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 48.4% |
| 31 | Claude Opus 4.8Qwen3.8 官方模型卡对照表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 48.4% |
| 32 | GPT-5.6 SolAtria Dawn Preview official card / report Table 1 · published comparison run · model snapshots and reasoning tiers not specified unless stated · official OpenCode harness · main split / 65 tasks · 7,200-second limit · two attempts · Grok-4.3 judge · mean task score· Reported by Shanghai AI Laboratory· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 45.4% |
| 33 | GPT-5.6 Solmax · 65 tasks · official evaluation code · OpenCode · file-aware rubric grader · mean rubric score· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 45.4% |
| 34 | GPT-5.6 SolQwen3.8 官方模型卡对照表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 45.4% |
| 35 | GPT 5.5xhigh · OpenCode · Grok 4.3 judge · official JobBench results· Reported by Meta, Meta Superintelligence Labs· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 38.3% |
| 36 | Qwen 3.8 27B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 33.4% |
| 37 | Qwen 3.8 27BQwen3.8-27B model card · published setting· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 33.4% |
| 38 | Qwen3.7 MaxQwen3.8 官方模型卡对照表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 31.3% |
| 39 | Gemini 3.1 Pro Previewhigh · OpenCode · Grok 4.3 judge · official JobBench results· Reported by Meta, Meta Superintelligence Labs· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 15.9% |
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 |
|---|---|
| Muse Spark (version unspecified)Meta Model API · highest effort · version unspecified · OpenCode · Grok 4.3 judge · official JobBench results· Reported by Meta, Meta Superintelligence Labs· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 17% |
Data credits: 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. · Data by LLM Stats