GPT-6.1 Sol
Key numbers
- Panel score
- 58rank 6 of 11
- Cost
- $0.83API list price
- Wall time
- 14 min
- Model time
- 13 min91% of wall time
- tok/s
- 36.1
- Tokens
- 4.0MOutput: 27k
- Cache hit
- 98%
- API calls
- 47
- Tool calls
- 46
- Renderer
- Intel UHD 630 (GPU)~60 ms / frame
- H1–H6
- Pass
- Cost per point
- $0.014
How the run ended
Finished cleanly with a video that passes every hard requirement and its own report. The audio track is silent, which the spec allows.
Blind panel
Scores of the three judges in every panel this video was part of.
| Panel | Letter | Anthropic | OpenAI | Median | Spread | Rank | |
|---|---|---|---|---|---|---|---|
| 14 Nov 2026 11 videos | G | 58 | 58 | 60 | 58 | 2 | 6 / 11 |
| 13 Nov 2026 10 videos | C | 60 | 58 | 69 | 60 | 11 | 5 / 10 |
| 11 Nov 2026 9 videos | D | 58 | 58 | 67 | 58 | 9 | 4 / 9 |
Sub-scores and the judges’ one-line notes from the panel of 14 Nov 2026.
| judge | Landmarks / 25 | City look / 25 | Camera & drone / 30 | Life / 10 | Overall / 10 | Total |
|---|---|---|---|---|---|---|
| Anthropic | 14 | 13 | 18 | 8 | 5 | 58 |
| OpenAI | 11 | 12 | 23 | 7 | 5 | 58 |
| 18 | 17 | 14 | 6 | 5 | 60 |
Judges’ notes
- AnthropicDrone always visible with tram, birds and crowds, but flat untextured buildings and a few blocked frames.
- OpenAIThe prominent drone and busy bridge read clearly, though boxlike landmarks and overhead obstructions limit the result.
- GooglePleasant warm roofs and visible lower-center drone, but heavily penalized for pressing into the solid bridge-tower wall at 36.2 s and a wall band at 10.9 s.
Time and speed
Wall time includes rendering and encoding. Model time is the sum of all model calls.
- Model generating
- Tools & waiting
| Wall time | 14 min |
|---|---|
| Model time | 13 min |
| Tools & waiting | 1.3 min |
| Model share of wall time | 91% |
| tok/s | 36.1 |
| Model calls | 47 |
Tokens and cost
Tokens summed over every model response. Tokens multiplied by the provider’s price.
| Uncached input | 87,145 |
|---|---|
| Cache read | 3,846,656 |
| Cache write | 0 |
| Output | 27,421of which 5,807 reasoning |
| Total | 3,961,222 |
| Cache-hit ratio | 97.8% |
| Cost | $0.83API list price |
| Output tokens per point | 473 |
| Points per $ | 69.9 |
Tools used
Tool calls by name, as the harness logged them. Names differ between harnesses.
Rendering
| Renderer | Intel UHD 630 (GPU) |
|---|---|
| Final render | 2.0 min |
| Per frame | 60 ms (approximate) |
Hard requirements H1–H6
Checked independently from the delivered MP4 with ffprobe.
| H1 | h264 High yuv420p | Pass | |
|---|---|---|---|
| H1 | faststart | Pass | |
| H2 | 1080×1920 SAR1:1 | Pass | 1080x1920 SAR 1:1 |
| H3 | 30 fps | Pass | 30 fps |
| H4 | 55-60 s | Pass | 58.000 s |
| H5 | peak≤ 20Mbps, <200MB | Pass | peak 14.46 Mbps over any 1 s, 67.9 MB |
| H6 | AAC-LC 48k stereo | Pass |
From the agent’s report
Short, verbatim quotes from the REPORT.md the agent wrote itself.
The style is intentionally low-poly and architectural rather than photorealistic.
Contact sheet and route map
