Kimi K3.1 Pro takes on the pelican-on-a-bicycle 3D test: beats Opus 5, trails 5.5
blaind0 · reddit · 2026-10-24
A Redditor tested Kimi K3.1 Pro (default max thinking, 160k token context, Pi.dev harness) on the classic 'pelican riding a bicycle' 3D scene prompt and compared it to Opus 5.5's launch result. Verdict: better than Opus 5, but still a sizable gap to Opus 5.5 and Astra 6; more thinking budget might have helped. The output includes detailed rigging and pedal animation.
More from Models
- Moonshot removed cyber-related training data from open-weight Kimi K3.1 Pro — nrehiew_ · 2026-10-24
- User says Opus silently downgraded to 4.8 mid-task, then told him to talk to people — natesiggard · 2026-10-24
- Physical-engineering benchmark: Claude Opus 5.5 tops at 48.3%, Grok 4.7 trails at 30.1% — teortaxesTex · 2026-10-24
- Suspicious new GitHub accounts reference Qwen4 in Alibaba T-Head repo PRs — brainExploded99 · 2026-10-24
- Kimi's $19 coding plan roasted: 5-hour limit hit before a single meaningful line of code — teortaxesTex · 2026-10-24
- Kimi's new model reportedly 30x costlier to serve than V4.1 Flash, sparking skepticism — teortaxesTex · 2026-10-24