Darth Vader 3D comparison
The author shares Darth Vader 3D models from Sonnet 5.5 and GPT-6 Sol.
@AiBattle_RECENTLY UPDATED / (SHANGHAI TIME)
CHEER LIN / ARCHIVE
Browse 87 public Claude Sonnet 5.5 use cases, evaluations, and reports of limitations. Check the tasks, outcomes, and conditions with original sources, creator accounts, and tool roles preserved.
USE CASE MAP / 01
Cases, evaluations, and limitations explicitly attributed to Sonnet 5.5 are retained with promotional disclosures, companion tools, failures, and unfinished work. Results come from public demos or creator accounts and have not been individually reproduced by this site. Record counts do not represent independent successful projects.
HOW TO USE / 02
Browse by task, then check the original inputs, toolchain, test conditions, and reported outcomes to choose a method to test yourself.
Filter by category or search titles, methods, and creators to find demos, evaluations, and limitations.
Confirm the source explicitly uses Sonnet 5.5 and identify the steps performed by external tools or other models.
Open the creator’s post for context, disclosures, inputs, revisions, failures, and unfinished work.
Connect through EvoLink, prepare the inputs and tools shown in the case, and record cost, time, and actual outcomes.
REAL CREATOR CASES / 03
Up to five cases per row, with fewer columns on narrower screens. Select a creator to open the original case; images and videos load only when needed.
The author shares Darth Vader 3D models from Sonnet 5.5 and GPT-6 Sol.
@AiBattle_At High effort, the author reuses an earlier Opus prompt for a 30-second product reel with music, real assets and elaborate motion, then shares Sonnet's result.
@HO_BAThe author reports a four-minute JavaScript rendition of Goldilocks and the Three Bears with character voices, music and sound effects; the audio tools are unspecified.
@WinterArc2125Using a skill based on Clawd Animation Base, the author splits the music video into 14 parallel chapters. Opus 5.5 writes the story and Sonnet 5.5 animates it.
@LLMJunky
CASE 05TutorialThe author promotes a seven-page workflow report covering a lead plus five subagents, blind review, diff and cost gates. The 120-worker scale and efficiency figures are the author's claims.
@Argona0xThe author reports a roughly 30-minute xHigh run using tools and references to make a lyric motion video for their own song from one prompt.
@original_ngv
CASE 07EvaluationThe author asks Sonnet 5.5 Max to create the Mona Lisa in SVG and references an earlier Opus attempt.
@diegocabezas01BridgeMind's own promotional demo reports a one-pass zombie game at Max effort taking 49 minutes and $177, illustrating the task cost behind the showcase.
@bridgemindaiHiggsfield's own account presents a 40-v-40 shooter with airstrikes built using Sonnet 5.5 and its platform.
@higgsfield_aiThe prompt asks for a 3D BMW M5 in Three.js. The author explicitly reports one prompt and no edits before sharing the output.
@k2sbhaiThe author asks Sonnet 5.5 to make a video explaining its own appeal and shares the result. Its pricing and performance messaging remains promotional claims.
@NFT_ChenThe author reports early access to Sonnet 5.5 and an Age of Empires-style game they can patch themselves, with a multiplayer play link.
@ForwardEditorThe author reports a one-prompt Mario Kart game and calls it their best result on this test, a personal assessment of that run.
@matthewmilleraiUsing the same prompt, the author compares GPT-6 Astra Ultra and Sonnet 5.5 xHigh on a 3D Cerebras chip benchmark and discusses interactive learning.
@SPAC89The author makes an interactive plasma-globe desktop wallpaper with Sonnet 5.5 and Three.js; the quality comparison with Opus is personal opinion.
@chaseleantj
CASE 18EvaluationCognition's own launch post reports 64.4% on FrontierCode 1.1 Main and availability in Devin Desktop and CLI. The score is the platform's reported evaluation.
@cognition
CASE 19EvaluationThe author gives Sonnet 5.5 and Opus 5.5 the same prompt to build GTA Rio, a PES-style Grenal game and Squid Game, reporting no extra human prompts.
@oalanicolas
CASE 21EvaluationThe author asks Sonnet 5.5 and Opus 5.5 to draw self-portraits in different styles using JavaScript and shares their outputs.
@kickingkeysThe post presents a one-minute human-evolution animation whose every frame, the author says, was drawn in JavaScript by Sonnet 5.5.
@vibeevalThe author shares a Sonnet promotional reel made with code, including its music, and promotes a tutorial covering how seven videos were made.
@petergyangClaude's own account reshares @_re_pete's fall-foliage simulator comparing Sonnet 5 and 5.5, providing a second distribution source for the creator's work.
@claudeaiClaude's own account reports that @LEJ_limited supplied a 60×46-pixel emoji, requested a printable 3D model and had a print hours later. This records the post's account.
@claudeaiClaude's own account reshares @kevin_t_ngo's JavaScript dinosaur-history animations, reporting the same prompt for Sonnet 5 and 5.5 and a runtime under a minute.
@claudeaiClaude's own account reshares @forwardeditor's bouncing-ball physics tests, reporting the same prompt for Sonnet 5 and 5.5.
@claudeaiThe author reports a GIF animated in Python and rendered in Blender, made jointly with Opus 5.5 and Sonnet 5.5. Their individual roles are not specified.
@kevin_t_ngoThe author adapts GMI Cloud's Sakura engine into a late-1990s neighborhood using Opus 5.5 and Sonnet 5.5, reporting about one hour and $55–57 in tokens.
@AmirMushichThe author reports a one-prompt browser 3D tornado scene with a barn being torn apart and controls to steer the tornado.
@ivanainaiAt High effort, the author compares Tokyo voxel scenes: 49m05s and $9.23 for Sonnet 5.5 versus 21m50s and $4.85 for Sonnet 5, using their reported bills.
@AiBattle_
CASE 32TutorialThe author shares a Claude Projects configuration tutorial recommending Opus 5.5 as coordinator and Sonnet 5.5 for threads.
@dani_avila7With the same prompt and reference image for a Three.js Jeep, the author reports 1h25m and 6% weekly quota for Sonnet versus 35 minutes and 7% for Astra; detail judgments are subjective.
@maxt3chnoThe author gives Sonnet 5.5, Opus 5.5 and GPT-6 Astra one prompt each to build a 3D planet and shares the comparison; Sonnet uses Max effort.
@monokernThe author redesigns a personal website with Sonnet 5.5 and provides a site link and showcase, with a subjective assessment of design improvement.
@_tenZdhon_The post reports a cycling-pelican test: one roughly 68 KB SVG using SMIL without JavaScript, with linked wheel, pedal and chain motion plus limited feather detail.
@GKev1nThe author presents a Sonnet 5.5 Fortnite test, reporting one prompt and an output close to the original game, a judgment about this demo.
@ericmaskfr
CASE 38DemoOn a $20 subscription, the author builds and refines a full-stack app at Low effort, reporting nearly four hours and 86 iterations in that session.
@IndependentEcoThe author tests GPT-6 Sol and Sonnet 5.5 with a landing-page prompt and shares a visual comparison. The preference for Sonnet's polish is subjective.
@blueemi99The author compares Low through Max on one motion brief, assigning subjective grades from C+ to A. xHigh is judged most tightly beat-synced; Max is ambitious but silent.
@cherry_mx_redsUnder the same prompt and limits, Sonnet 5.5 and GPT-6 Sol each control a block-placing bot to build a colossal Minecraft dragon, with the author's preferred result.
@siyabuilt
CASE 42DemoThe author presents a Sonnet 5.5 C++ skate game described as self-contained and dependency-free.
@bijanbowenAt Max effort, the author's New York scene test produces an automated city tour with parks, traffic and named landmarks, compared with an earlier Sonnet 5 run of the same prompt.
@aipulseda1lyThe official tutorial provides a Medium-effort SDK request, a /claude-api command for migrating from Sonnet 5, and the /model sonnet switch in Claude Code.
@ClaudeDevsThe author supplies six CLAUDE.md rules: Jev selects a profile, the host fixes model and effort, Sonnet handles scoped coding, and completion requires checks. The fourfold speedup is self-reported.
@Av1dlive
CASE 46EvaluationIn the author's nine-defect static code test, Medium changes 3/9 defect signatures while High and xHigh reach 9/9. High takes 9m38s and $2.17; xHigh takes 50m56s and $13.85.
@SKatalystAIThe author requests a Three.js CoD Zombies clone with breakable windows, rebuildable barriers, Mystery Box, perks and weapons, reporting code-generated textures, sounds and icons.
@intheworldofaiThe author compares Sonnet 5.5 Max and Opus 5.5 Max on the cycling-pelican test and invites readers to inspect the results.
@wanxuboThe author builds a four-floor puzzle game, reporting roughly an hour to finish in friends-and-family testing. Voices use ElevenLabs; a persistent stalking enemy was omitted.
@The_AlexThe author gives Sonnet 5.5 and Opus 5.5 the same prompt for a mechanical watch assembling itself in midair and publishes the outputs.
@karankendreBridgeBench compares the same lava-lamp prompt: Sonnet at $0.47 and 5m44s, Opus at $1.27 and 9m35s. These figures describe that reported test.
@bridgebenchShowing 48 / 87
BUILD WITH EVOLINK / 04
Access Claude Sonnet 5.5 through EvoLink. Use the cases to prepare inputs and required tools, start with a defined task, and record cost, time, and whether the result meets your requirements.