Pika API
Listed at https://pika.art/api
💰 Token Pricing
| Type | Price | Note |
|---|---|---|
| Input | 会员制 $10/月 + 首月 $10 额度。按模型批发价按量计费：GPT Image 2.0 $22.50/1M token；Seedance 2.0 R2V 480P $0.0903/秒；MiniMax H3 T2V 2K $0.13/秒 | per million tokens |
| Output | GPT Image 2.0 图像 $22.50/1M token；Seedream 5.0 Pro 2K $0.09/张；Nano Banana Pro $0.134/张；音频按分钟/请求/千字符；视频按秒 | per million tokens |
🤖 Supported Models (120)
✨ Pros
- ✓API Club aggregates 100+ frontier Gen-Media models: Seedance / Kling 3.0 / Veo 3.1 / MiniMax H3 / GPT Image 2 / Claude Fable 5 under one API
- ✓Wholesale pricing: frequently 20-88% below fal.ai / Runway API (GPT Image 2.0 at $22.50 vs fal's $30.00)
- ✓Transparent no-hidden-markup pricing: $10/mo membership + pay-as-you-go, model-by-model prices published with a savings calculator
- ✓Agent-native: one onboarding prompt brings it into Claude Code / Codex / Cursor / Lovable / ChatGPT
- ✓Enterprise committed-use discounts up to 70% off, higher concurrency, reserved capacity, dedicated Slack, custom models
- ✓All API outputs licensed for commercial use; you retain ownership of prompts and inputs
- ✓99.9% uptime; shared Slack support + dedicated devrel team for every API customer
⚠️ Cons
- ×No free API tier; the $10/mo membership is the entry cost (first-month $10 credit roughly offsets it)
- ×China access requires proxy (US-hosted, no mainland edge nodes)
- ×Member wholesale rates include a 5% platform fee; not guaranteed lowest on every single model (Pika is transparent about this)
- ×Aggregates third-party models, subject to upstream version churn — older models can be swapped out
- ×Primarily video/image/audio generation; LLM calls are a supplementary feature, not the core
- ×Video prices nest by resolution/duration (480P-4K), so cost estimation requires per-tier price-page lookups
🎯 Best For
Budget-conscious image/video/audio production that wants to A/B or route across many Gen-Media models; developers who want to generate media directly inside Claude Code / Cursor-class agents
💰 Pricing & Plans
| Item | Pika API Club price | Competitive reference | Savings |
|---|---|---|---|
| Membership | $10/mo (+ $10 first-month credit) | — | — |
| GPT Image 2.0 (text-to-image) | $22.50 / 1M tokens | fal.ai $30.00 | 25% off |
| Seedance 2.0 R2V 480P (video) | $0.0903 / sec | Runway API $0.36 / sec | 75% off |
| MiniMax H3 T2V 2K (video) | $0.13 / sec | fal.ai $0.26 / sec | 50% off |
| Seedream 5.0 Pro 2K (image) | $0.09 / image | fal.ai $0.135 | 33% off |
| Nano Banana Pro 1K (image) | ~$0.134 / image | fal.ai $0.15 | 33% off |
| Seed Audio TTS | $0.15 / min | fal.ai $0.1875 | 20% off |
| Minimax Music (T2A) | $0.09 / req | fal.ai $0.15 | 40% off |
| ElevenLabs TTS v2 | $0.105 / 1k chars | fal.ai $0.10* | varies |
| Enterprise (committed use) | up to 70% off volume | — | — |
🔧 API & Developer Experience
- •Payment & Auth: API Club membership is $10/mo; you generate an API key from the Pika dashboard and pay member wholesale rates per generation. New members get a $10 first-month credit that roughly covers the entry fee. No opaque per-call markup — a flat membership plus visible, model-by-model usage pricing.
- •Agent-Native Onboarding: The API is designed to be legible to AI agents, not just human devs reading docs. You can copy one onboarding prompt directly into Claude Code, Codex, Cursor, Lovable, Replit, or ChatGPT and start generating media immediately — no multi-page integration guide.
- •Model Breadth (100+): One API routes to video (Seedance 2.5, Kling 3.0, Veo 3.1, MiniMax H3, Pika 2.5, Wan 3.0, FLUX 3, Hailuo, Happyhorse), image (GPT Image 2, Seedream 5.0 Pro, Nano Banana Pro, Ideogram 4.0, Recraft 4.1, Grok Imagine), audio (ElevenLabs Suite, MiniMax Music/Speech, Seed Audio), and LLMs (Claude Fable 5, GPT-5.6 Sol, Gemini 3.x, DeepSeek V4 Pro, Kimi K3, GLM-5.x).
- •Output Licensing: All API outputs are licensed for commercial use under the Pika API agreement; you retain ownership of your prompts and inputs.
- •Enterprise Controls: Committed-use volume discounts up to 70% off, higher concurrency limits, reserved capacity during demand spikes, a dedicated solutions engineer, private Slack channel, tailored SLAs, and custom models tuned by the Pika Research team.
- •Reliability & Support: 99.9% uptime across all models; every API customer gets a shared Slack channel with the developer-relations team, plus email support (api@pika.art) with ~1 business-day turnaround.
- •Transparent Cost Calc: The pricing page publishes every model's member rate alongside fal.ai and Runway API reference prices, with a savings calculator — no need to reverse-engineer blended per-call costs.
🧩 API Club: Wholesale Aggregation Model
Pika's API Club is a wholesale buying club rather than a conventional usage-priced gateway. Instead of baking a margin into every API call the way many Gen-Media aggregators do, Pika negotiates model rates downward on members' behalf and collects a flat $10/mo membership plus pay-as-you-go member pricing. The published price table shows Pika at up to 88% below other aggregators on headline models — GPT Image 2.0 at $22.50/1M tokens vs fal.ai's $30.00 (25% off), Seedance 2.0 480P at $0.0903/sec vs Runway API's $0.36/sec (75% off). Pika is explicit that they cannot guarantee being cheapest on every model and keeps a public comparison page with a savings calculator rather than hidden rebates, which is what makes the model transparent. The membership economics self-fund quickly: generating three 15-second Seedance 2.0 clips saves more than the $10 fee versus Runway Dev pricing. Enterprise tiers layer committed-use discounts (up to 70% off), reserved capacity, and custom models on top of the same wholesale engine. A 5% platform fee is included in listed member prices (recorded 8/23/26); outputs are licensed for commercial use. Prices verified 2026-09-04.
🌐 Regional Availability & Latency
Pika Labs is a US-based company; the Pika API and its generation jobs run on Pika's cloud infrastructure, which is not served from a mainland-China edge node. Developers in mainland China reach the API only through a stable overseas proxy or VPN, with latency dependent on the generation job itself (video/image jobs are asynchronous) rather than on a per-token round-trip. For teams outside mainland China the API is fast and geographically broad — Pika routes against multiple upstream Gen-Media providers (Alibaba, ByteDance, Google, MiniMax, Moonshot, Anthropic, OpenAI, xAI), so a given request may be executed by whichever provider currently has capacity, and the pricing stays uniform regardless of the executing region. There is no China-domestic pricing tier, no Alipay / WeChat Pay top-up, and no stated mainland data-residency option. Commercial-use licensing applies globally. Teams shipping media features to mainland users should plan an overseas-egress hop or route their LLM/media inference through a China-accessible provider while keeping Pika for non-mainland traffic.