RECENTLY UPDATED / (SHANGHAI TIME)
LIN YUEJI / ARCHIVE
Explore 116 public cases showing Claude Fable 5.1 across games, frontend work, 3D, coding evaluations, media production, and workflows—with the evidence limits kept visible.
USE CASE MAP / 01
This collection preserves the key result, evidence type, creator, and original source for 116 public cases so you can quickly spot repeatable methods.
HOW TO USE / 02
Narrow by task, verify context at the original source, then reproduce the most useful path through EvoLink.
Start with coding, agents, creative work, evaluation, or another task family.
Use titles, methods, or creators to find the closest case to your current goal.
Select the creator to review the full demo, limits, context, and original explanation.
Choose a method worth testing, connect to the model, and turn it into your workflow.
REAL CREATOR CASES / 03
Up to four cases per row. Select a creator to open the original case; images and videos load only when needed.
116 cases
CASE 07DemoA single prompt produced 100 varied HTML experiments. The source post reports strong frontend quality and few broken files, but this collection treats the full set as one experiment rather than 100 independently validated cases.
CASE 12DemoThe creator shows a multiplayer survival game made from one prompt with cooking, crafting, hunting, and dinosaur taming. Multiplayer reliability and the claimed 20+ hours of gameplay were not independently verified.
CASE 33DemoThe same creator shows an Ultracode subway FPS and separately reports that Fable 5.1 spent two hours on its design.md. The demo confirms visible game output, not an efficient planning process.
CASE 48DemoBridgeMind demonstrates a kart-racing game reportedly generated in one shot. The video shows a playable-looking result, while the complete prompt and acceptance criteria were not disclosed.
CASE 66DemoAn early-access user shares a retro spaceship game inspired by FTL as an example of long-running work requiring judgment and taste. This is a disclosed early-access demo, not an independent blind test.
CASE 15DemoStarting from a property-lot photo, the team reports that Fable 5.1 designed a house, rendered it, and produced a cinematic walkthrough through code. The tools and dimensional accuracy were not disclosed.
CASE 53DemoThe creator reports producing the motion video in 19 minutes with two changes. Because revisions were made, this should not be described as an unedited one-shot result.
CASE 56EvaluationAn AI/ML API demo compares five continuous Three.js scenes at $5.69 for Fable 5.1 versus $0.88 for GPT-5.6 Sol. Fable shows richer detail but also ships a pool without water; the publisher sells API access.
CASE 62EvaluationThe Fable 5.1 post says the same prompt produced a markedly different ship-in-a-bottle visualization. The older-model post identifies the object, but its run configuration should not be attributed to Fable 5.1.
CASE 19ExampleThe author reports that Fable 5.1 reimplemented an old paper, corrected approximations, derived another solvable model, and extended future work over seven hours. The results were not independently reproduced; the GPU-kernel claim in the same post refers to Mythos, not Fable 5.1.
CASE 79ExampleVals AI reports eliciting a proposed decoding of the 373-year-old Cyphral Distich. This is a first-person test report; the decoding has not been independently validated and should not be presented as an accepted solution.
CASE 24ExampleThe creator recommends running a prompt audit across existing skills and reports that it finds redundant rules. The post describes an active batch run, not a completed audit with a published change list.
CASE 80DemoAt Max thinking, the author generated a pelican SVG and then animated it in a separate step, reporting a $3.30 cost. The disclosed workflow is multi-step, not a one-shot animation.
CASE 81EvaluationThe author reports that Fable 5.1 deleted 33,000 lines in a code-deletion benchmark, versus 1,000–2,000 for earlier models. This collection does not independently verify that the deletion was correct.
CASE 82DemoThe author shows a publicly accessible frontend animation and praises the model’s consistency. The complete prompt, revision count, and acceptance checklist were not disclosed.
CASE 83DemoThe post shows generated Chinese cursive calligraphy. The visual result is public, but its character forms and calligraphic quality have not been professionally reviewed.
CASE 84EvaluationUsing the same prompt as an earlier-model run, the author generated a sunken neo-Gothic tower as a Twigl shader. The comparison reflects the creator’s own visual assessment.
CASE 85EvaluationThe same game task compares UI, mechanics, stability, and cost across Fable 5.1 and Kimi K3. The author prefers Fable’s polish while noting Kimi’s lower cost and fewer gameplay jitters; the judgment is subjective.
CASE 86DemoA short public prompt produced a browser shooter with recoil, aiming, weapon weight, and three weapon classes. A playable link is provided in the replies, but long-session reliability was not fully tested.
CASE 87EvaluationFrom one floor plan and the same prompt, Fable 5.1 reportedly took 111 minutes and produced 14 stills plus a 44-second walkthrough. More output does not by itself establish greater architectural accuracy.
CASE 88EvaluationThe post compares Fable 5.1 and GPT-5.6 Sol on a similar anime-scene video task. It shows both results but does not provide a shared scoring rubric.
CASE 89DemoThe creator says a fantasy ascent of Mount Fuji was generated entirely in code from one prompt. A visual result is shown, while the full run and revision history are not disclosed.
CASE 90EvaluationOne prompt produced terrain, inventory, caves, TNT, and block destruction in a single HTML file. The author tested the explosion crater, but not long-running behavior across the full game.
CASE 91DemoThe author shows a one-shot tool that cuts a puppy out of a photo and turns it into a digital sticker. The complete code, cost, and acceptance criteria were not disclosed.
CASE 92EvaluationIn a 20-minute visual-control sandbox, Fable 5.1 left the starting area in four minutes, one minute slower than Fable 5. None of the tested models found the required destination.
CASE 93DemoA third party reports that a friend generated a Mario-style game in one turn with music, sound effects, fireballs, mushrooms, and character mechanics. The post is a retelling, not a direct operator report.
CASE 94EvaluationBox reports early testing across finance, technology, and public-sector document tasks with concrete analytical examples. This is a vendor-run early evaluation, not an independent customer testimonial.
CASE 95EvaluationThe author reports a feature review costing $15.20 with Fable 5.1 versus $20.76 with Fable 5. The review scope and quality of findings were not standardized or disclosed.
CASE 96DemoThe creator reports one prompt generated 2.4 million voxels, 1,500 soldiers, 160 NPCs, destructible buildings, and a dragon. The published video shows the world, but its scale and gameplay were not independently verified.
CASE 97DemoThe post shows a generated Chinese ink-style scroll. The prompt, number of iterations, and aesthetic evaluation criteria were not disclosed.
CASE 98TutorialThe post defines a 2–4 player LAN security exercise where each participant runs and defends a local web service while probing peers. It specifies game rules but does not publish a complete run result.
CASE 99DemoThe author reports generating a photorealistic Three.js landscape in 26 minutes. The visible result is public, while the full prompt and cost are not disclosed.
CASE 100EvaluationA simple prompt compares Fable 5.1 High and Fable 5 High on a Three.js helicopter model. The post shows both outputs without an objective scoring rubric.
CASE 101DemoThe author says Fable 5.1 generated onboarding for a new macOS app in one shot. The artifact is shown, but interaction and release acceptance were not disclosed.
CASE 102DemoUsing a full product brief, the author reorganized BaoCut v2 around agents, history, recent projects, and subtitle translation entry points. The result remains a prototype with no production usage data.
CASE 103DemoThe author reports an 18-minute, code-only walkable world plus an automatically exported 15-second video. Three correction rounds were required, so it was not a one-shot result.
CASE 104DemoTwo prompts produced a Three.js game with cinematic and gameplay views, steering, speed controls, missions, and a public deployment. The collection did not independently play-test it.
CASE 105EvaluationUsing the same one-shot prompt, the author compares Fable 5.1 with Opus 5 on an ocean sandbox and highlights rain and scene detail. The quality judgment and cost interpretation are the author’s own.
CASE 106ExampleAn audit moved YouTube, email, and deal polling to a self-hosted automation server, waking the bot through one webhook only when an event arrives. The post does not disclose measured billing savings.
CASE 107EvaluationThe author reports no collisions on their hardest multi-constraint benchmark. The rubric and reproducible scoring details were not published.
CASE 108EvaluationThe post shows Fable 5.1’s result on a Boeing 747 generation benchmark and promises prompt details in replies. This collection does not independently score the model.
CASE 109EvaluationTwo prompts reportedly produced a 3D game in about 20 minutes, with the model finding and fixing many bugs before yielding. The author did not provide an independent pass rate.
CASE 110DemoThe post shows a game-like World War I voxel simulator. It does not disclose the prompt, runtime, or a validation of its mechanics.
CASE 111EvaluationIn one short-story test, the author preferred Fable 5’s structure and prose, noting that Fable 5.1 used too many short sentences. One run is not enough for a general conclusion.
CASE 112EvaluationAcross sticky-ball, capybara-surfing, and dumpling-dodge games, Atomic Agent reports total cost falling from $7.65 to $7.08. This is a platform-run comparison, not an independent benchmark.
CASE 113EvaluationA platform-run comparison reports Fable 5.1 used 1.2M tokens, $123.98, and 59.8 minutes for a racing game—more than Fable 5—while producing more detailed assets and smoother play.
CASE 114BenchmarkOn two repositories with 105 planted bugs, the author reports Fable 5.1 fixed 43, versus 42 for GPT-5.6 Sol. This collection did not rerun the benchmark.
CASE 115IntegrationThe workflow combines a 3D-generation MCP, image and 3D models, Computer Use, Blender, and a texture-painting app. The outcome cannot be attributed to Fable 5.1 alone.
Showing 48 / 116
BUILD WITH EVOLINK / 04
Connect to the model through EvoLink and build from directions already demonstrated in public.
