In this guide 7 sections
Higgsfield Soul 2.0 turns a text prompt into a photo for 0.12 credits, and in our 2026-10-27 test it did not keep one face. We made 11 images of our fictional, adult, AI-generated test character: 8 Soul 2.0 preset runs, 2 Soul 2.0 reference runs and 1 Soul Cinema run. By eye, none matched our reference face. Total charge: 1.32 credits.
Disclosure: we may earn a commission if you subscribe through our Higgsfield links, each marked “(commission link)”. It does not change what we test; see the affiliate disclosure.
Part of the Higgsfield templates catalog.
What is Higgsfield Soul? (tested 2026-10-27)
Higgsfield Soul is Higgsfield’s photorealistic image model; we tested Soul 2.0 on 2026-10-27. The vendor calls Soul a “hyper-realistic, fashion-grade AI photo model with 50+ aesthetic presets” (Higgsfield Soul page, checked 2026-10-27). Our session listed 33 presets. The pricing page listed Soul 2.0 at 0.12 credits per image and the earlier Soul at 0.25 (Higgsfield pricing page, checked 2026-10-20).
We tested fashion photos of one person via the Higgsfield MCP tool on a Plus plan. Jobs came back labeled text2image_soul_v2, a record we draw no conclusion from.
How to use Soul 2.0: the presets and prompts we ran
In our 2026-10-27 runs, the Soul 2.0 preset outranked the prompt: one of four presets replaced our scene. Two prompts, four presets, one try each, at 3:4 and 2k, via MCP rather than the web app.
- Choose Soul 2.0 in the Higgsfield image generator: Soul on Higgsfield (commission link).
- Pick one preset or one reference image, not both.
- Describe the person: age, hair, glasses, clothes, pose, place. Skip lighting words when a preset sets the look.
- Check the credit cost, then generate.
Prompt 1 (waist-up, café):
Tested prompt
The same text as in the recipe card above. Copy it from the card
Prompt 2 (full body, crosswalk):
Tested prompt
Full-body fashion photo of an original fictional adult woman, about 29 years old, shoulder-length curly copper-auburn hair, light freckles, round tortoiseshell glasses, small silver hoop earrings, wearing an olive-green bomber jacket, a plain white t-shirt, dark straight-leg jeans and white sneakers, walking across a quiet city crosswalk, whole body in frame from head to feet. Ordinary person, not a celebrity, no text or logos.
One run per cell, 0.12 credits each. Faces are judged by eye against our reference image G0, with no face-matching tool.
Preset | Prompt 1 (café) | Prompt 2 (full body) |
|---|---|---|
Flash editorial | Café and cup as asked; different face | Crosswalk, one foot cut off; different face |
Editorial street style | Scene replaced: striding across a street, cropped top | Stride across the road, cropped top again, adds a phone |
Disposable camera | Café and cup as asked; adds a septum ring and red nails | Dusk flash snapshot; one foot cut off |
Warm vivid film | Café, but chin on hand and the cup on the table | Crosswalk in warm film color; one foot cut off; hair darker brown |
All 8 kept the glasses and hoops we described, yet character consistency did not hold: each run drew a different woman. One run per cell says nothing about other prompts.
Run H13-a1, Flash editorial: as asked, with a different face.
Run H13-a2, Editorial street style: same prompt, moved to a street.
Run H13-b4, Warm vivid film: warm film color; one foot leaves the bottom edge.
Presets and a reference image cannot be combined
The server refused a preset plus our reference image at the quote step: “Soul style_id cannot be combined with image references (medias)”. No job, no charge. The message adds that a soul_id can be used with a style. We have not trained a Soul ID (an identity trained on your photos), so that pairing is not tested.
A reference image copied the backdrop, not the face
As a control, runs H13-ref1 and H13-ref2 used both prompts with no preset and G0 as the reference. Both returned waist-up portraits on G0’s gray studio backdrop, ignoring the café and the crosswalk. By eye, the face did not match: a longer face and wavy, lighter hair on prompt 1; a broad toothy smile and large hoops on prompt 2.
Run H13-ref1: G0’s backdrop and framing, a different face.
The reference steered composition, not identity. G0 is a Soul 2.0 image from 2026-10-21 (4 variants, 0.48 credits); see the test character page.
What it costs: 0.12 credits per image
Soul 2.0 charged 0.12 credits per image on all 10 of our runs on 2026-10-27, at 2k and 3:4. The 1.5k quote was also 0.12, and every quote matched the charge. With Soul Cinema, the 11 images cost 1.32 credits: about $0.065 at the Plus monthly price ($59 for 1,200 credits, checked 2026-10-20). Per image, that is about $0.0059, or $0.0047 billed annually ($47). Plans are on our Higgsfield pricing guide.
Sent all at once, 3 of the 11 were refused with “Rate limit reached: max 8 concurrent job(s) on plus (monthly) plan.” No job was created and nothing was charged. All 3 went through when resubmitted unchanged about 50 seconds later.
Soul 2.0 vs Soul Cinema
On one shared input on 2026-10-27, Soul Cinema came nearer to our reference face than Soul 2.0, but did not match it. Run U5 used prompt 1 and the G0 reference, like H13-ref1. The job record reads soul_cinematic; our quote and charge were 0.12 credits.
Soul Cinema returned a tight close-up on a pale green backdrop with warm side light, ignoring the café, the cup and the framing. By eye, the round face, freckles and broad smile were nearer to G0 than in any other image. The hair came out darker and frizzier. One run, no face tool.
Run U5, Soul Cinema with the G0 reference: nearer to G0, café ignored.
That 0.12 is our own quote only. Our 2026-10-20 snapshot of the pricing page has no Soul Cinema row, so its list price is not verified as of 2026-10-27.
Is Soul free on any plan?
Not on our account: on 2026-10-27 our Plus account paid 0.12 credits for every Soul 2.0 image. The MCP catalog marks Soul 2.0 as supporting unlimited use; our account showed it as unavailable. Our 2026-10-20 snapshot of the pricing page quoted “5,000 FREE Soul 2.0 generations” on the Plus card. It read “Not included” in the Free column of the Soul 2.0 row. Both are not verified as of 2026-10-27: the live page showed no price text to our fetch.
Higgsfield’s help center says generations through MCP, CLI, Canvas or Supercomputer “always deduct credits” (Higgsfield help center, checked 2026-10-27), so our MCP runs were all paid.
For free images, start with what is free on Higgsfield. Pay for Soul 2.0 if you want preset looks at 0.12 credits and accept a new face per image. For one recurring face, our test found no support.
Recreate on Higgsfield (commission link)
FAQ
What is Higgsfield Soul?
Higgsfield’s photo model, which the vendor calls hyper-realistic and fashion-grade. Soul 2.0 charged us 0.12 credits per image on 2026-10-27.
Soul 2.0 vs Soul Cinema?
One run each on one prompt and reference, 2026-10-27: Soul Cinema came nearer to the reference face, Soul 2.0 copied its backdrop. Both charged us 0.12 credits; our pricing snapshot has no Soul Cinema row.
Which Higgsfield model is free?
Not verified as of 2026-10-27. Our 2026-10-20 snapshot of the pricing page showed “5,000 FREE Soul 2.0 generations” on the Plus card; our account had none on 2026-10-27.
Sources checked
- Our test: runs H13-a1 to H13-b4, H13-ref1, H13-ref2 and U5, Higgsfield Plus via MCP, 2026-10-27; G0 from 2026-10-21. Method: how we test.
- Higgsfield, Soul page. Checked 2026-10-27.
- Higgsfield, pricing page: credits per image, plans, Plus card, Free column. Checked 2026-10-20; not verified as of 2026-10-27.
- Higgsfield, help center: Unlimited models. Checked 2026-10-27.
- Our calculation: credits × $0.0492 (Plus monthly, $59 / 1,200) or × $0.0392 (annual, $47 / 1,200).
