DeepSeek V4.1 Flash (Max)
- Cost
- $0.43
- Time Spent
- 1h 34m
- Harness
- OpenCode · v2.0.32
- Model version
- 1017
- Tokens in / out
- 1.3M / 73.7K
- Date
Accuracy
12.5%
How closely this build matches the reference model.
3D view
Loading 3D model…
Drag the divider to reveal the build · drag the model to orbit · scroll to zoom
Model views
front
Loading front comparison images
back
Loading back comparison images
left
Loading left comparison images
right
Loading right comparison images
Details
- Parts placed
- 146
- Reference parts
- 339
- Parts matched
- 54
- Positions matched
- 0
- Orientations matched
- 0
- Surplus parts
- 0
- Match
- 0.053
- Unbound parts
- 26
- Bad parts
- 26
- Structural
- 0.822
- Car shape
- 1
Other builds of this benchmarkOther AI models, coding harnesses or reasoning effort levels on this same benchmark. Each row represents one setup and shows its build with the highest reference-accuracy score. Rows run from the highest score to the lowest. This build's setup is excluded.
| Harness · model | Accuracy | Cost | Time Spent |
|---|---|---|---|
| Codex · GPT 6 Astra (Ultra) | 78.1% | $33.06 | 1h 25m |
| Claude Code · Claude Sonnet 5.5 (Max) | 58.8% | $2.89 | 8h 0m |
| Claude Code · Claude Opus 5.5 (Xhigh) | 50.3% | $88.83 | 2h 30m |
| Claude Code · Claude Sonnet 5.5 (Xhigh) | 48.4% | $146.09 | 3h 36m |
| Codex · GPT 6.1 Sol (Xhigh) | 38.6% | $16.32 | 3h 58m |
| Codex · GPT 6.1 Sol (Ultra) | 33.6% | $10.18 | 2h 58m |
| Antigravity · Gemini 4 Argon (High) | 22.7% | $18.76 | 4h 0m |
| Claude Code · Claude Fable 5.1 (Xhigh) | 16.0% | $112.67 | 3h 6m |
| Codex · GPT 6 Luna (Max) | 11.8% | $1.97 | 3h 46m |
| Grok Build · Grok 4.6 (Xhigh) | 11.1% | $14.48 | 1h 6m |
| OpenCode · DeepSeek V4.1 Flash (Max) | 10.4% | $0.64 | 3h 3m |
| Grok Build · Grok 4.7 (Xhigh) | 10.0% | $38.02 | 53m 40s |
| Kimi Code · Kimi K3 (Max) | 10.0% | $64.44 | 4h 0m |
