Interface em português. Títulos, descrições, prompts e documentos dos autores são preservados no idioma da fonte ou na versão em inglês do arquivo.
SVG estáticoRegistro da comunidadeSimon Willison tested Claude Sonnet 5.5: the xhigh run took 41 seconds and cost about 5.74 cents, producing an SVG with a mostly correct frame, legs, pedals and chain, but a misshapen blue bicycle helmet. The max run spent 128,000 tokens thinking and failed to return an SVG.
Obra 3DRegistro da comunidadeSimon Willison asked the Claude Sonnet 5.5 free tier to build a WebGL HTML page rendering a three-dimensional pelican riding a bicycle, and published the runnable single-file page. This archive keeps a copy for its interactive demo; the page still loads Three.js from a CDN.
VídeoRegistro da comunidadeA community author ran a one-line HTML/SVG animation brief at xhigh and posted a recording of the result. The original prompt is preserved, but the post does not say whether the output was revised, so this is catalogued only as a self-reported community sample.
AnimaçãoRegistro da comunidadeA standalone HTML+SVG animation output submitted by Wang-Kaijun on Pelican Benchmark. Model attribution: MiMo v2.6 Pro (thinking level: Not sure, environment: chat interface). Contains native vector artwork and self-contained animation logic. Model attribution is source-reported and not independently authenticated.
AnimaçãoRegistro da comunidadeA standalone HTML+SVG animation output submitted by Wang-Kaijun on Pelican Benchmark. Model attribution: Kimi K2.8 Preview (Max) (thinking level: Max, environment: chat interface). Contains native vector artwork and self-contained animation logic. Model attribution is source-reported and not independently authenticated.
AnimaçãoRegistro da comunidadeA standalone HTML+SVG animation output submitted by Wang-Kaijun on Pelican Benchmark. Model attribution: Grok 4.7 (High) (thinking level: High, environment: chat interface). Contains native vector artwork and self-contained animation logic. Model attribution is source-reported and not independently authenticated.
AnimaçãoRegistro da comunidadeA standalone HTML+SVG animation output submitted by Wang-Kaijun on Pelican Benchmark. Model attribution: GPT-6 Astra (XHigh) (thinking level: XHigh, environment: chat interface). Contains native vector artwork and self-contained animation logic. Model attribution is source-reported and not independently authenticated.
AnimaçãoRegistro da comunidadeA standalone HTML+SVG animation output submitted by Wang-Kaijun on Pelican Benchmark. Model attribution: GLM-5.3-Flash (Max) (thinking level: Max, environment: chat interface). Contains native vector artwork and self-contained animation logic. Model attribution is source-reported and not independently authenticated.
AnimaçãoRegistro da comunidadeA standalone HTML+SVG animation output submitted by Wang-Kaijun on Pelican Benchmark. Model attribution: Gemini 3.8 Flash (High) (thinking level: High, environment: chat interface). Contains native vector artwork and self-contained animation logic. Model attribution is source-reported and not independently authenticated.
AnimaçãoRegistro da comunidadeA standalone HTML+SVG animation output submitted by Wang-Kaijun on Pelican Benchmark. Model attribution: DeepSeek-V4.1-Flash (Max) (thinking level: Max, environment: chat interface). Contains native vector artwork and self-contained animation logic. Model attribution is source-reported and not independently authenticated.
AnimaçãoRegistro da comunidadeA standalone HTML+SVG animation output submitted by Wang-Kaijun on Pelican Benchmark. Model attribution: Claude Opus 5 (High) (thinking level: High, environment: chat interface). Contains native vector artwork and self-contained animation logic. Model attribution is source-reported and not independently authenticated.
AnimaçãoRegistro da comunidadeThe published process includes frontend/SVG references, browser rendering, visual checks and repeated edits, not one-shot generation. The author reports interruption when credits ran out. The posted animation is preserved without claiming a completed run. The date is public posting, not an authenticated generation date; model attribution is source-reported.
Obra 3DRegistro da comunidadeThe publisher posts a five-second two-model comparison for a looping Blender cycling prompt. The post and video panel consistently label this output GPT-6 Sol. This page shows only that model’s work region; all 150 decoded frames match the original region pixel for pixel. The complete comparison is retained in the collapsed attachments. A real nonblank frame at one second is the cover, not another work. Promotional superiority claims are not museum conclusions. Full operations, settings and iterations are unpublished; one-shot generation is not asserted. Model labels and attribution are source-reported, not independently authenticated. The date is the original post’s UTC publication day, 2026-09-27, not a known generation date; Beijing time may display the following day.
Obra 3DRegistro da comunidadeThe publisher posts a five-second two-model comparison for a looping Blender cycling prompt. The post and video panel consistently label this output Opus 5.5. This page shows only that model’s work region; all 150 decoded frames match the original region pixel for pixel. The complete comparison is retained in the collapsed attachments. A real nonblank frame at one second is the cover, not another work. Promotional superiority claims are not museum conclusions. Full operations, settings and iterations are unpublished; one-shot generation is not asserted. Model labels and attribution are source-reported, not independently authenticated. The date is the original post’s UTC publication day, 2026-09-27, not a known generation date; Beijing time may display the following day.
AnimaçãoRegistro da comunidadeThe original lower-left panel says xhigh - one prompt, matching the body’s xhigh label. The author uses pi and custom skills. Medium also received use the skills; no separate unnudged version is published, so none is invented. Four actual outputs count separately; the original medium represents the timeline, not the most polished later refinement. Model labels and attribution remain source-reported, not independently authenticated. The date is verifiable public posting, not inferred generation. Author costs, times and opinions are not museum scores or capability findings.
AnimaçãoRegistro da comunidadeThe lower-right panel explicitly says medium + detailed prompt. The author describes further detailed prompting and checking its own screenshots in the existing medium session, three prompts deep in total: not a new reasoning level, independent blind attempt or one-shot result. The complete six-second final-version recording is preserved; its frames are not extra works. Model labels and attribution remain source-reported, not independently authenticated. The date is verifiable public posting, not inferred generation. Author costs, times and opinions are not museum scores or capability findings.
AnimaçãoRegistro da comunidadeThe original upper-left panel says medium - one prompt + use skills, matching the body’s medium label. The author uses pi and custom skills. Medium also received use the skills; no separate unnudged version is published, so none is invented. Four actual outputs count separately; the original medium represents the timeline, not the most polished later refinement. Model labels and attribution remain source-reported, not independently authenticated. The date is verifiable public posting, not inferred generation. Author costs, times and opinions are not museum scores or capability findings.
AnimaçãoRegistro da comunidadeThe original upper-right panel says high - one prompt, matching the body’s high label. The author uses pi and custom skills. Medium also received use the skills; no separate unnudged version is published, so none is invented. Four actual outputs count separately; the original medium represents the timeline, not the most polished later refinement. Model labels and attribution remain source-reported, not independently authenticated. The date is verifiable public posting, not inferred generation. Author costs, times and opinions are not museum scores or capability findings.
AnimaçãoRegistro da comunidadeA standalone HTML+SVG animation output submitted by Anonymous on Pelican Benchmark. Model attribution: Hunyuan 4 (hy4) (Tencent Hunyuan 4). Contains native vector artwork and self-contained animation logic. Model attribution is source-reported and not independently authenticated.
AnimaçãoRegistro da comunidadeA Brazilian developer generated an HTML animation in one shot using an English prompt.
AnimaçãoRegistro da comunidadeA Turkish-language discussion of a LuminaBench comparison, questioning whether it really used Gemini 4.
AnimaçãoRegistro da comunidadeA custom demoscene instruction set fitting IK and a cycling pelican into 4 KB.
AnimaçãoRegistro da comunidadeThe author rated it above Grok 4.7. The post advertised free access in OpenCode until October 10.
AnimaçãoRegistro da comunidadeGenerated in 58 seconds, with an awkward-looking result; 19,000 views.
