# Command: `linksite:model-bench`

**Description**: Run the multi-model refresh-content bench over pinned fixtures
**Status**: completed
**Started**: 2026-07-08 18:07:23 | **Ended**: 2026-07-08 20:17:02 | **Duration**: 2h 9m
**Jobs**: 0 dispatched / 0 completed / 0 failed

---

## Command Output

18:07:23 [INFO] Lock acquired for command: linksite:model-bench

18:07:23 [INFO] → refresh × jp-1 × aisubscription

18:13:54 [INFO] → refresh × jp-1 × anthropic/claude-opus-4.8

18:46:54 [INFO] → refresh × jp-1 × anthropic/claude-sonnet-5

19:15:33 [INFO] → refresh × jp-1 × deepseek/deepseek-v3.2

19:47:02 [INFO] → refresh × jp-1 × google/gemini-3.1-pro-preview

19:57:00 [INFO] → refresh × jp-1 × moonshotai/kimi-k2.6

20:17:02 [INFO] winners → score: aisubscription · cost: aisubscription · speed: aisubscription · overall: aisubscription

20:17:02 [INFO] Artifacts: /home/guilherme/webapps/kokos-laravel2/storage/app/bench-results/6a4e91dbc14a7c53880942a3

20:17:02 [INFO] Run id: 6a4e91dbc14a7c53880942a3

20:17:02 [INFO] Lock released for linksite:model-bench

---

## Jobs (0)

*No jobs dispatched*

*Exported at 2026-08-05T07:31:53+00:00*