Measured harness ledgerPublic result
DeepSeek V4 FlashHigh-voxel-density Jungle Temple diorama — DeepSeek V4 Flash Max
Create a complete inspectable Blender voxel diorama of a jungle temple at a fixed 0.05-unit voxel resolution through the Blender MCP workflow.
Max reasoningHeadline result
- Workflow cost
- $0.06
- Wall-clock
- 21m 39.4s wall-clock
- Processed tokens
- 5.27M processed
- Record state
- partial_token_timing_and_artifact_ledger
Public summary
DeepSeek V4 Flash Max partial_token_timing_and_artifact_ledger ledger: 21m 39.4s wall-clock, 5.27M processed, and $0.06 API-equivalent usage accounting, not an itemized subscription cash charge.
Run identity and stack
- Result ID: jungle-temple-deepseek-v4-flash-max
- Technical model: DeepSeek V4 Flash
- Provider: OpenCode Go
- Client: OpenCode
- Stack: OpenCode Go
- Stack: OpenCode
- Stack: Technical model/configuration: DeepSeek V4 Flash
- Stack: Blender MCP
- Stack: 0.05-unit voxel grid
- Stack: Inspectable .blend artifact
Cost basis
- Visible output and reasoning are both output-priced. OpenCode Go subscription pricing is not divided into a marginal cash cost for this run.
Primary artifact integrity
- Kind: blender-scene
- Path: artifacts/jungle-temple-deepseek-v4-flash-max/jungle_temple_diorama.blend
- SHA-256: c620ddb89d4ff46509d20aa97e134b1c1ce4e784375cf5becec8d6d29c97657f
Validation evidence
- Result: not_run
Recorded caveats
- Wall-clock is end-to-end workflow latency including tool execution, Blender MCP calls, and idle gaps; it is not model-only compute.
- One OpenCode agentic turn can contain many billable model requests.
- Cache-read tokens are deeply discounted, so total processed tokens overstate API-equivalent cost.
- The public record excludes private OpenCode session/export data; their source hashes and stable benchmark-window hash are retained.
- The `.blend` scene and builder source are hash-bound, but no independent Blender inspection, fresh render, local FPS measurement, RTX Pro 6000 capture, or blind-evaluation record was supplied.
Visible evidence gaps
- independent Blender load and render validation
- local FPS measurement
- final RTX Pro 6000 1080p60 render or capture metadata
- blind-evaluation record
Public result only
This result keeps its public summary and evidence, but it does not currently have a matching Builder test with prompts, projects, Harness workflows or skills.