Harness comparisonAI Coding
Pelican cycling a coast road, three models in 3D — Claude Code (Claude Opus 5.5) vs OpenAI Codex (GPT-6 Astra) vs DeepSeek Harness (DeepSeek V4.1 Flash 1017)
Scoreboard3
- Claude Opus 5.5AnthropicClaude Code
- Time
- 1 h 15 min
- Effort High
- GPT-6 AstraOpenAIOpenAI Codex
- Time
- 11 min
- Effort High
- medium confidence
- Written “GPT-6.1 Astra” in the text; the video labels it GPT-6 Astra.
- DeepSeek V4.1 Flash 1017DeepSeekDeepSeek Harness
- Time
- 56 min
- Effort High
Pelican cycling a coast road, three models in 3DThe post doesn’t quote the prompt wording.
The poster says Opus made a golden-hour coast road with a detailed pelican and is still far ahead; DeepSeek’s sea light stood out and Astra is clean cartoon.
@0xarthurzs’s view (paraphrased)
Claude Opus 5.5 in Claude Code, GPT-6 Astra in Codex CLI and DeepSeek V4.1 Flash 1017 in DeepSeek Harness get the same plain prompt for a 3D Three.js film of a pelican riding a bicycle.
Each side pairs a coding harness with its own model, so harness and model differences are mixed. More Claude Code vs OpenAI Codex. More OpenAI Codex vs DeepSeek Harness. More Claude Code vs DeepSeek Harness. All harness comparisons.
The post, itemised
- Task
- Pelican cycling a coast road, three models in 3D
- Category
- 3D & Shaders, Animations/SVG
- Mode
- Not stated
- Posted
- Video length
- 9 s
- Likes
- 0
- Views
- 121
- Bookmarks
- 0
- Sponsored
- No
Cost & time, as stated in the postAs stated: Claude Opus 5.5 75 min, 41.7M tokens; DeepSeek V4.1 Flash 1017 56 min, 22.56M tokens; GPT-6 Astra 11 min, 2.22M tokens.
Related
Posted by @0xarthurzs on X. aivs.tv claims no ownership: it hosts a copy for fast playback and links to the post on X. Request removal
