Pika and HiggsField are all-in-one providers, like WaveSpeed: one key each unlocks a set of models from many vendors. Pika brings a large live catalogue of image, video and audio models plus chat models. HiggsField brings 14 image and video models, including its own Cinema Studio.
Both keys go in the same place: click API Keys in the app header and open the All-in-One tab. See Configuring API Keys.
Pika
Add the key
- Create a key on Pika’s developer site (the Get an API key from Pika link on the key’s row opens it).
- Paste it into Pika API Key on the All-in-One tab and click Save.
The app then reads Pika’s catalogue and refreshes it once a day, so new Pika models appear without an app update. In the model picker, Pika models carry a yellow P badge and end with [Pika]; click the Pika chip at the top of the picker to list only them.
What the catalogue includes
One Pika key loads every image, video, audio and utility model Pika publishes, each filed under its category in the right panel. Alongside Pika’s own models (Pika 2.5, Pikaffects, Pikaswaps, Pikadditions and the audio line below), the catalogue includes hosted models from other vendors, such as Seedance, Kling, Veo, Gemini, GPT Image, Grok Imagine, Seedream, Ideogram, Recraft, Wan, Hailuo, MiniMax, FLUX 3, ElevenLabs, Lyria, Sonilo, Sync and Topaz. The exact list follows Pika’s catalogue.
| Panel | Categories with Pika models |
|---|---|
| Generate Image | Text to Image, Image to Image (Edit), Image Upscale |
| Generate Video | Text to Video, Image to Video, References to Video, Omni, Video to Video, Video Extend, Avatar, Motion Control, Video Upscale |
| Generate Voice | Text to Speech, Text to Audio, Speech to Speech, Text to Music, Transcribe (Audio), Video to Audio, Video to Music, Voice Isolation, Voice Dubbing |
Pika’s transcription models live in Generate Voice under Transcribe (Audio). They are not used for voice input: dictation needs an OpenAI or WaveSpeed key.
Duration and other settings on the bar
Every Pika model that has a length shows it as a Duration dropdown (the clock pill) on the bar next to the model picker, not in Advanced Settings. Aspect ratio and resolution also sit on the bar as pills.
- Wide ranges are offered in steps, and the list always keeps the model’s shortest and longest length.
- The default is 5 s, or 30 s for music.
- Video extensions open on auto, which measures your source clip.
- A length that only describes an input (for example, how long the source audio is) is measured from the media you picked, so you don’t set it.
Options that Pika leaves for the vendor to decide open on Default. Keep Default to let the vendor choose, or pick a value from the list.
Pika’s audio line
| Model | Category | What it does |
|---|---|---|
| Pika Speech | Text to Speech | Speaks your script in one of 76 voices from Voice Preset, or in a voice taken from a Reference Audio recording. |
| Pika SFX | Text to Audio | A sound effect from a prompt, 1–20 s (default 10 s). |
| Pika Music | Text to Music | Music from a prompt, with optional lyrics, in three modes: text to music, voice-conditioned and cover. 10–360 s (default 30 s). |
| Pika Soundtrack | Video to Audio | Writes a synchronised soundtrack for a video and returns the same video with its audio replaced. The picture and length stay untouched. |
The Video Editor’s transcript and voice-clip panels don’t offer Pika’s voice models, because those panels re-speak a take in its own voice. Pika voices stay available in Generate Voice, where any voice is a valid answer. See Audio, Speech and Music Generation.
Chat models
The Pika key also adds 14 chat models to Board Chat and the Prompt Enhancer, each ending with [Pika]: Claude Fable 5, Claude Opus 5, Claude Sonnet 5, GPT-5.5, GPT-5.6 Luna, GPT-5.6 Sol, GPT-5.6 Terra, Gemini 3.1 Pro, Gemini 3.5 Flash-Lite, Gemini 3.6 Flash, DeepSeek V4 Flash, DeepSeek V4 Pro, Kimi K3 and GLM-5.2. They read text only, so use another provider’s model when you want the chat to look at images.
Pricing
- The price next to Generate comes from Pika’s published pricing for the settings you picked, including the resolution tier and the duration (hover it to see Exact price from Pika).
- Some hosted models are billed by tokens. Their price reads API priced, because no fixed price can be shown in advance; check your usage in Pika’s dashboard.
HiggsField
Add the key
HiggsField issues a key id and a key secret. Paste both into HiggsField API Key (KEY_ID:KEY_SECRET) on the All-in-One tab as one value, joined by a colon: KEY_ID:KEY_SECRET. Without the colon, generating fails with an error that explains the format. The row’s link (Get API key with up to 50% discount) opens a HiggsField sign-up with a discount.
HiggsField models carry an H badge and end with [HiggsField] in the model picker.
The 14 models
| Model | Category | What it does |
|---|---|---|
| MiniMax H3 Omni | Omni | 2K reference-to-video from up to 9 images, 3 videos and 3 audio clips. |
| Kling O3 Video Reference | Omni | Video-guided generation from one reference clip, a prompt and up to 4 stills. |
| Seedance 2.5 Omni | Omni | Reference-to-video with image, video and audio references, up to 30 s. |
| Cinema Studio 4.0 | Omni | Directed shot generation with a director’s kit: Genre, Era, Pacing, Lighting, Camera, Lens, Aperture, Camera Movement (33 moves) and Color Palette (50 palettes). |
| Seedance 2.5 Video Edit | Video to Video | Rewrites a clip from a written instruction, with optional stills and audio. |
| Kling O3 Video Edit | Video to Video | Prompt-driven edit of one clip, up to 4K, with up to 4 reference images. |
| Genjutsu Motion Transfer | Video to Video | Transfers the motion of a clip onto the characters in your stills. |
| Genjutsu Object Swap | Video to Video | Swaps an object or character in a clip for the one in your stills. |
| Kling O3 First-Last Frame | First-Last Frame | Animates between a first and a last frame, up to 15 s and 4K. |
| Kling O3 Image Reference | Image to Video | Animates up to 4 reference stills, with optional first and last frames. |
| Seedance 2.5 Video Extend | Video Extend | Continues an existing clip by up to 30 s. |
| Kling V3 Motion Control Pro | Motion Control | Drives a character still with the motion of a reference clip. |
| Soul | Text to Image | HiggsField’s text-to-image model at 720p or 1080p, with optional Soul styles and 1 or 4 results. |
| Ideogram 4.0 | Image to Image (Edit) | Image generation and editing with strong text rendering; the input image is optional. |
HiggsField has no chat, audio or transcription models.
How HiggsField runs
- Reference images, clips and audio you attach are uploaded to HiggsField before the job starts, so full-length clips work as references.
- The price next to Generate is HiggsField’s own estimate for exactly the settings you picked (hover it to see Exact price from HiggsField). If the estimate is unavailable, the panel falls back to the app’s estimate from published rates.
- Options left on Default let HiggsField apply its own default.
Where to go next
- Configuring API Keys — where the All-in-One keys go.
- Video Generation — using Omni, Video to Video and the other video categories.
- Audio, Speech and Music Generation — the Generate Voice panel, where Pika’s audio line lives.
- Using WaveSpeed — the third all-in-one provider.
- Custom AI Models — add a provider the app doesn’t include.
Leave a Reply