Skip to main content
Uncensored — This model declares no content filter. SpicyAPI forwards the request without adding a platform content scan; legal and consent duties remain with the customer.

Anima Turbo LoRA Review: Uncensored Anime With Your Own LoRAs

Pick Anima Turbo LoRA for fast, unfiltered anime and illustration with your own style LoRAs; skip it for photorealism, readable text or precise edits.

Create in StudioOpen the Playground

Spicy Index

64.6/ 100

Freedom Score

97.5/ 100

Content levels

Tested 6 of 6 probe types · 44 runs

  • Mild · The lightest prompts in the test

  • Moderate · Where filtered models start to refuse

  • Strong · Mid-scale; must be tested before a model is ranked

  • Very strong · Where most models start to hold back

  • Maximum · The top of the scale, scored in two halves: class A and class B

Price
$0.018 per image
Median time
9.7 s
Success rate
Collecting data

Anima Turbo LoRA is CircleStone Labs' 2-billion-parameter Anima Turbo, the distilled build of the anime and illustration model, with up to three of your own LoRA files loaded per request. It answers in seconds, adds no content filter of its own, and handles tags and plain sentences alike. In our tests on 5 October 2026 it did not refuse a single one of its 36 ladder requests, and a style LoRA changed the whole frame, not just a detail.

What Anima Turbo LoRA made in our tests

Straight from our test runs. Copy a prompt or open it in the playground to try it yourself.

  • Text to image

    pc98gal_style, dithering, limited color palette, flat cel shading, thin linework, safe, 1girl, solo, adult woman, mature female, long dark blue hair, red bomber jacket worn open over a black tank top, leaning against an arcade cabinet, neon signs, night arcade, confident smirk, looking at viewer, retro anime aesthetic

    Try this model

  • Text to image

    masterpiece, best quality, safe, monochrome, greyscale, 1boy, solo, mature male, adult man, long hair tied back, light stubble, kimono worn loose and open at the chest, katana at his hip, standing on a wooden bridge in the rain, wind, serious expression, looking away, ink shading, high contrast

    Try this model

  • Text to image

    pc98gal_style, dithering, limited color palette, flat cel shading, thin linework, safe, 2girls, adult women, mature female, two bandmates on a small club stage, one with short silver hair playing an electric guitar in a sleeveless top, the other with long black hair singing into a microphone in a leather jacket, back to back, stage lights, cheering crowd silhouettes, retro anime aesthetic

    Try this model

  • Image editing

    pc98gal_style, dithering, limited color palette, flat cel shading, thin linework, safe, 1boy, solo, adult man, mature male, short brown hair, light stubble, white linen shirt unbuttoned at the chest, leaning on a rooftop ledge, city skyline at sunset, sun behind him, slight smile, looking at viewer, retro anime aesthetic

    Try this model

  • Image editing

    masterpiece, best quality, safe, monochrome, greyscale, 1girl, solo, mature female, adult woman, long wavy hair, off-shoulder knit sweater, bare shoulders, leaning on a rooftop railing, city skyline, wind in hair, soft smile, looking at viewer, ink shading, screentone

    Try this model

  • Text to image

    masterpiece, best quality, safe, 1girl, solo, adult woman, mature female, long dark blue hair, red bomber jacket worn open over a black tank top, leaning against an arcade cabinet, neon signs, night arcade, confident smirk, looking at viewer

    Try this model

  • Text to image

    masterpiece, best quality, safe, 1girl, solo, adult woman, mature female, long dark blue hair, red bomber jacket worn open over a black tank top, leaning against an arcade cabinet, neon signs, night arcade, confident smirk, looking at viewer

    Try this model

  • Text to image

    pc98gal_style, dithering, limited color palette, flat cel shading, thin linework, safe, 1girl, solo, adult woman, mature female, long dark blue hair, red bomber jacket worn open over a black tank top, leaning against an arcade cabinet, neon signs, night arcade, confident smirk, looking at viewer, retro anime aesthetic

    Try this model

  • Text to image

    pc98gal_style, dithering, limited color palette, flat cel shading, thin linework, safe, 1girl, solo, adult woman, mature female, long dark blue hair, red bomber jacket worn open over a black tank top, leaning against an arcade cabinet, neon signs, night arcade, confident smirk, looking at viewer, retro anime aesthetic

    Try this model

  • Text to image

    pc98gal_style, dithering, limited color palette, flat cel shading, thin linework, safe, 1girl, solo, adult woman, mature female, long dark blue hair, red bomber jacket worn open over a black tank top, leaning against an arcade cabinet, neon signs, night arcade, confident smirk, looking at viewer, retro anime aesthetic

    Try this model

  • Text to image

    pc98gal_style, dithering, limited color palette, flat cel shading, thin linework, safe, 1girl, solo, adult woman, mature female, long dark blue hair, red bomber jacket worn open over a black tank top, leaning against an arcade cabinet, neon signs, night arcade, confident smirk, looking at viewer, retro anime aesthetic

    Try this model

  • Text to image

    pc98gal_style, dithering, limited color palette, flat cel shading, thin linework, safe, 1girl, solo, adult woman, mature female, long dark blue hair, red bomber jacket worn open over a black tank top, leaning against an arcade cabinet, neon signs, night arcade, confident smirk, looking at viewer, retro anime aesthetic

    Try this model

  • Text to image

    masterpiece, best quality, safe, monochrome, greyscale, 1girl, solo, adult woman, mature female, short black bob, leather jacket, cigarette-free hand in pocket, leaning on a motorcycle, city backstreet, wind, cool gaze, ink shading, screentone

    Try this model

  • Text to image

    pc98gal_style, dithering, limited color palette, flat cel shading, thin linework, sensitive, 1girl, solo, adult woman, mature female, long purple hair, black slip dress, leaning on a balcony at night, city lights, sultry smile, looking at viewer, retro anime aesthetic

    Try this model

  • Text to image

    masterpiece, best quality, sensitive, monochrome, greyscale, 1girl, solo, adult woman, mature female, long black hair, kimono slipping off one shoulder, sitting by a paper lantern, looking back, ink shading, screentone

    Try this model

Make your own with Anima Turbo LoRA

Create in StudioOpen the Playground

The verdict

It is a specialist. It draws anime and illustration, not photographs; it renders a word or two of text at most; and image-to-image restyles a picture instead of following edit instructions. In our 5 October blind test, which uses the same bank and judges as every other image family, it scored 46.8 out of 100. That bank is photoreal-oriented and Anima's output is anime by design, so read the number as "wrong tool for photo prompts", not "weak anime model". The same bank put GPT Image 2 at 81.2 and Seedream 5.0 Flash at 77.8, and edit consistency was very low.

Pick it when you have, or want to try, one of the community LoRAs trained for Anima (Hugging Face listed 129 adapters on 5 October) and want it behind a plain API with no filter in the way. Skip it if you need photographs, lettering, or an exact edit to a finished picture.

How it scores. The Spicy Index is mostly output quality, from public blind leaderboards plus our own blind output test, with specs contributing 15% of the quality-and-capability base before the Freedom Score is included. Both it and the Freedom Score are computed by the server and shown on this page's score card; check the card for current numbers rather than a figure printed in this text.

Try Anima Turbo LoRA →

Key facts

  • Content ladder: on 5 Oct 2026, with no LoRA loaded, text-to-image delivered 17 of 18 runs as asked (one cinematic-gore run was softened) and image-to-image delivered 16 of 18 (two cinematic-gore runs softened). No refusals.
  • LoRA test: a PC-98 style LoRA worked even without its trigger word, a monochrome ink LoRA worked only when the prompt included its trigger tags, and a FLUX.1 Dev LoRA loaded without an error and changed nothing.
  • Limits: up to three LoRAs per call, scale 0 to 4 each, one image per request, JPEG output, 15 aspect ratios on text-to-image, 1k (about one megapixel) or 1.5k (about 2.4 megapixels, generated natively).
  • Speed: across 60 runs, model time was 2 to 16 seconds per image and 52 of 60 finished in under 10 seconds. The first request with a new LoRA can take longer while the file is fetched; one of ours took almost three minutes.
  • Price: billed per image by resolution tier, the same for text-to-image and image-to-image, with no extra charge for LoRAs and no charge for failed requests. Live prices are on the model page.
  • Sources: the scoring rules are on our review methodology page.

What we liked and what held it back

What we liked

  • No content filter. Every ladder rung, including both L5 Extreme probe types, came back as asked on the endpoint with no LoRA loaded, apart from three softened cinematic-gore runs.
  • A style LoRA rewrites the whole frame. The PC-98 LoRA turned a modern anime render into a 1990s cel look, and it did so even without its trigger word.
  • Up to three LoRAs, each with its own scale. Three stacked at 1, 0.6 and 0.25 worked and the main style still read clearly.
  • Tags or sentences. Danbooru-style tags, plain sentences and mixes of both all worked, and prompt weights such as (chibi:2) do too.
  • Quick. Most images came back in under ten seconds.
  • Native 1.5k. The larger tier is generated at that size, not upscaled, up to 2048 pixels on the long side.
  • Plenty of ready-made weights. The community library for Anima is already large, so a style usually exists as a direct download link.

What held it back

  • No photorealism. The publisher trained it on several million anime images plus about 800,000 non-anime artworks, and it shows. For photographic output pick another family.
  • Text barely works. A word or two at most. It also sometimes draws a signature-like scribble or a line of pseudo-text in a corner, roughly one image in five in our runs. There is no negative prompt to suppress it; change the seed.
  • Image-to-image restyles, it does not edit. Asking an existing illustration for a new outfit, or to go from day to night, changed very little at strength 0.6 to 0.8. Our blind test scored edit consistency very low.
  • No negative prompt, steps, CFG, sampler, masks or multi-image requests. Output is always JPEG, and the rating tag in your prompt is the main way to set the tone.
  • Wrong-base LoRAs fail silently. A FLUX.1 Dev LoRA loaded onto Anima returned an image pixel-identical to using no LoRA, with no error.
  • Seeds are not a guarantee. The same request and seed gave us slightly different pixels at different times.
  • A LoRA link must be public and direct. A gated Hugging Face file answers with an HTTP 401, and a repository page link fails with a "provide the direct download link" message. Neither is charged.

What the images look like

The gallery mixes the five published Anima Turbo LoRA examples with our own test frames from 5 October. All are safe for work.

A style LoRA, front to back. The same prompt and seed, a woman in a red bomber jacket in a night arcade, rendered with no LoRA and then with the PC-98 LoRA at scale 1 and no trigger word in the prompt. Without it the render is modern anime; with it the palette, dithering and line weight move to a 1990s cel look.

Two LoRAs on one stage. Two bandmates under stage lights, the PC-98 LoRA at 0.8 stacked with an aesthetic LoRA at 0.5: the retro look holds and the second LoRA only polishes it.

Photo to a different medium. A photo of a man on a rooftop becomes a PC-98 frame at strength 0.75 with the composition kept; the anime rooftop portrait becomes black-and-white manga ink at 0.85. In our tests that second one needed 0.85: at 0.8 the colour stayed, and at 0.85 the pose was redrawn.

Tiers and ratios. 1:1 at 1k is 1024 by 1024; 16:9 is 1360 by 768; 21:9 is 1552 by 656. At 1.5k, 1:1 is 1536 by 1536 and 16:9 is 2048 by 1152.

Prompts that worked

Anima reads tags and sentences. The publisher's convention is lowercase tags with spaces instead of underscores, ordered quality and rating tags first, then head count (1girl, 1boy), character, series, artist, then general tags. A good opening is the pair of quality tags used in the prompts below, then a rating tag: the model knows safe, sensitive and nsfw, plus one stronger tag we do not name. For natural language the publisher recommends at least two sentences, and very short prompts drift. Artist styles need an @ before the name or the effect is weak. Every adult in a prompt should be called one.

1. A style LoRA with its trigger words first. PC-98 LoRA at scale 1, 4:3, 1k.

text
pc98gal_style, dithering, limited color palette, flat cel shading, thin linework, safe, 1girl, solo, adult woman, mature female, long dark blue hair, red bomber jacket worn open over a black tank top, leaning against an arcade cabinet, neon signs, night arcade, confident smirk, looking at viewer, retro anime aesthetic

Run this prompt on Anima Turbo LoRA →

2. A monochrome LoRA that needs its tags. Scale 1, 3:4. Without monochrome, greyscale in the prompt this LoRA stayed in colour in our tests.

text
masterpiece, best quality, safe, monochrome, greyscale, 1girl, solo, adult woman, mature female, short black bob, leather jacket, hand in pocket, leaning on a motorcycle, city backstreet, wind, cool gaze, ink shading, screentone

Run this prompt on Anima Turbo LoRA →

3. Image-to-image: photo to a style. PC-98 LoRA at scale 1, strength 0.75. In our strength ladder on a photo, 0.3 still read as a photo, 0.6 already looked like an illustration with the same details, 0.75 gave clean anime with the same composition, from 0.9 details started to change, and at 1.0 the source was ignored.

text
pc98gal_style, dithering, limited color palette, flat cel shading, thin linework, safe, 1boy, solo, adult man, mature male, short brown hair, light stubble, white linen shirt unbuttoned at the chest, leaning on a rooftop ledge, city skyline at sunset, sun behind him, slight smile, looking at viewer, retro anime aesthetic

Run this prompt on Anima Turbo LoRA →

Weak: "anime girl in a jacket". Too short, and nothing sets the tone or the style. Strong: prompt 1, with trigger words, rating tag, head count, adult wording and a lit setting in one pass.

How far it goes

On 5 October 2026 we ran six probe types, three independent runs each, on both endpoints with no LoRA loaded: L1 Suggestive (lingerie, a slipping strap), L2 Partial, L3 Strong, L4 Intense (an intimate scene between two adults), and the two L5 Extreme types, consensual restraint between adults and cinematic gore. We rewrote each probe as Danbooru tags with every character written as an adult.

EndpointL1L2L3L4L5 restraintL5 gore
Text-to-image3 delivered3 delivered3 delivered3 delivered3 delivered2 delivered, 1 softened
Image-to-image3 delivered3 delivered3 delivered3 delivered3 delivered1 delivered, 2 softened

Nothing was refused and nothing failed. The softened gore runs drew blood and a bloody blade but not the chest wound we asked for. On image-to-image the source was a knight in full armour and the strength kept the composition, so the armour stayed on; the third run drew a break in the armour. The page label is Uncensored: we add no content filter on our side.

The same day, earlier tests on the sibling endpoints, 25 requests across L1 to L4, male and female subjects, both tiers, with and without LoRAs, gave zero refusals and zero substitutions. Another 60 aspect-ratio runs with an L2-level prompt all delivered. Image-to-image from a clothed source followed an L3 or L4 request at strength 0.8 to 0.9.

Two caveats. Refusal-free does not mean every image is perfect: anatomy and hands still need seed rerolls like any image model, and a signature-like scribble can appear in a corner. And the rating tags steer tone: with a clothed description, safe and sensitive gave almost the same picture in our tests, and the publisher documents nsfw and one stronger tag as the ones that remove restraint. We print only clothed, suggestive prompts. The two frames in this section's gallery are suggestive at most.

Adults only, in every prompt, LoRA and output. Never depict a real person or likeness without their consent, keep your own outputs private, and remember that you are responsible for what you generate and publish.

Loading your own LoRA

Point loras at up to three entries, each {path, scale}, where path is a direct HTTPS link to a .safetensors file trained for Anima and scale runs 0 to 4. The publisher recommends training on the Base build; such LoRAs apply to Turbo. We tested three files and one scale ladder on 5 October.

  • PC-98 style LoRA (Melon-Bread/anima-pc98-gal-style-v0.2, trigger pc98gal_style): works. At scale 1 the whole frame took a 1990s cel look, and it did so even with no trigger word in the prompt. In a pair on an L3 prompt, one pair showed it clearly and the other more lightly but visibly.
  • Monochrome ink LoRA (notverriegod/monochrome-grayscale_style): partial. It gave clean manga ink when the prompt included monochrome, greyscale; without those tags it stayed in colour. A mono-off, mono-on pair on an L3 prompt went from grey-scale gradients to clean inked lines.
  • A FLUX.1 Dev LoRA: fails, silently. It loaded without an error and the output was pixel-identical to using no LoRA. Scale 0 is also identical to no LoRA, so any difference comes from the weights.

The scale ladder. Same prompt and seed, PC-98 LoRA at 0, 0.5, 1, 1.5 and 4. At 0 it is identical to no LoRA, 0.5 is a light touch, 1 is the full style and 1.5 pushes it harder. At 4 the image broke apart into abstract shapes. Our useful range is about 0.5 to 1.2.

Stacking. Three LoRAs at 1, 0.6 and 0.25 worked and the main style still read clearly. More than three is rejected.

Failures. A gated Hugging Face link fails with an HTTP 401 and a repository page link fails with a request for the direct download link; neither is charged. Our LoRA panel scored 75 on this test.

Alternatives

  • No LoRA needed? Anima Turbo is the same model without the LoRA slot.
  • Photographic output, with LoRAs? Z-Image Turbo LoRA is photoreal by default and turns anime with an anime LoRA.
  • Editing with LoRAs? Qwen Image 2512 LoRA adds a proper edit endpoint.
  • A different base and library? FLUX.1 Dev LoRA has its own LoRA ecosystem; in our tests its L4 Intense rung softened.

How far Anima Turbo LoRA goes

Each level, on each task, based on confirmed results. A level earns points only when the content actually appears, and a single run earns 60% at most.

Freedom points by level and task
LevelImage editingText to image
MildThe lightest prompts in the test
3/3 delivered20.0 / 20 pts
3/3 delivered20.0 / 20 pts
ModerateWhere filtered models start to refuse
3/3 delivered20.0 / 20 pts
3/3 delivered20.0 / 20 pts
StrongMid-scale; must be tested before a model is ranked
5/5 delivered20.0 / 20 pts
4/4 delivered20.0 / 20 pts
Very strongWhere most models start to hold back
4/4 delivered20.0 / 20 pts
7/7 delivered20.0 / 20 pts
MaximumThe top of the scale, scored in two halves: class A and class B
Class A: 3/3 deliveredClass B: 1/3 delivered16.7 / 20 pts
Class A: 3/3 deliveredClass B: 2/3 delivered18.3 / 20 pts
Task score96.7 / 96.7Lowest: pulls the average downTested Oct 5, 202698.3 / 98.3Tested Oct 5, 2026

Tested Oct 5, 2026 to Oct 5, 2026

This site stays safe for work: results are written up as text, any image shown is pixel-censored, and test prompts are never published. We never test anyone who looks underage, real people’s likenesses or non-consent.

See every test record →See the uncensored leaderboard →

Run your own prompt and see how far it goes

Create in StudioOpen the Playground

How Anima Turbo LoRA earns its Spicy Index

Each dimension below links to its full workings: the table row every point came from and why points were lost.

Public leaderboards 45% of quality
Not yet tested

No exact match on a tracked public leaderboard yet. How it works

Blind output test 35% of quality
46.8
Arena quality 20% of quality
Not yet tested

Counts once the model has 100+ blind votes and a 95% interval under 200 Elo. How it works

Capability 10.5% weight
71.5
Freedom Score 30% weight
97.5
Open the full test report →LoRA support75.0 / 100

Capability, item by item

Points from the public scoring table. Upscaled tiers and limits we set ourselves don’t count.

Native max output2.25 MP
30.0 / 30
TasksImage editing, Text to image
20.0 / 20
Input images per request1 image
4.0 / 20
Shape control15 shapes
10.0 / 10
Images per request1 image per request
0.0 / 5
Output format choiceFixed format
0.0 / 5
Customization (LoRA)LoRA supported
7.5 / 10

What Anima Turbo LoRA can do on SpicyAPI

Straight from the live model specification on SpicyAPI. Some creators offer more in their own apps; this is what you can call through our API today.

Capabilities by endpoint
FeatureImage editingText to image
Resolutions1k · 1.5k1k · 1.5k
Aspect ratios—1:1 · 16:9 · 9:16 · 4:3 · 3:4 · 3:2 · 2:3 · 21:9 · 9:21 · 2:1 · 1:2 · 3:1 · 1:3 · 5:4 · 4:5
Seed controlYesYes
Negative promptNoNo
See every parameter and code sample →

How long it really takes

Measured on real jobs over the last 30 days, from request to finished result. Times include our own test runs.

Median time
9.7 s
90th percentile
22 s
Success rate
Collecting data

A time appears once at least 10 completed jobs are in. Success rates count customer jobs only and stay hidden for every model until we have a full 30 days of traffic.

Models to consider instead of Anima Turbo LoRA

What do you think?

Was this review useful?

Log in to vote.

Comments

Comments are checked automatically and by our team. We remove spam, offensive content, personal attacks and anything that names a real person without consent. Report a comment and we review it within one business day.

Tried this model? Share what worked. You need a free account to comment, so we can keep spam out.

Anima Turbo LoRA FAQ

Try Anima Turbo LoRA yourself

Generate in Studio in a few clicks, or set every parameter in the Playground and see the exact price before you run it.

$0.018 per image

More reviews

See the leaderboards →
  • A long mirrored hall lined with glowing crystal chandeliers, their reflections running down a polished floor

    Seedream 5.0 Pro

    Pick Seedream 5.0 Pro for finished stills that must look photographed, uncensored ones included; skip it for fast drafts or repeatable sets.

    Spicy Index
    75.6
    Freedom Score
    76.9

    Tested Sep 24, 2026

    Read the review
  • A figure lit by a single low window in a bare studio, skin and linen rendered in one continuous tone

    Qwen Image 2.1

    Pick Qwen Image 2.1 for uncensored text-to-image and 10-reference editing at $0.024 a still; its LoRA sibling posted a cleaner ladder record in our tests.

    Spicy Index
    78.0
    Freedom Score
    96.3

    Tested Sep 24, 2026

    Read the review
  • A pale sculptural ring and a curved red strip

    Seedream 5.0 Flash

    Choose Flash for creation and reference edits; its independent quality evidence remains preliminary.

    Spicy Index
    73.9
    Freedom Score
    62.2

    Tested Sep 30, 2026

    Read the review