Replicate
Hosted API to run and deploy machine-learning models. Pay for use; official HTTP API, Python client, MCP server and agent skills.
Tags
Description
Commercial model-hosting API (replicate.com). Established provider gap fill, not a claim of a brand-new launch. Pricing page: you only pay for what you use. Some models are billed by hardware-time, others by input and output; estimates are on each model page. Hardware examples on that page (live page is authoritative): Nvidia T4 $0.000225/sec ($0.81/hr), L40S $0.000975/sec ($3.51/hr), A100 80GB $0.001400/sec ($5.04/hr), H100 $0.001525/sec ($5.49/hr). Public models: you pay only for active processing time; setup and idle are free, and you share a hardware pool (cold boots possible). Most private models and deployments bill setup, idle and active time; fast-booting fine-tunes bill active time only. Failed runs are not charged; canceling an official time-based model may still bill time already used. Billing docs: prepaid credit, or billed in arrears (previous month, sometimes early if usage first crosses certain limits). Free limits: you can run select models for free, then you are asked to set up billing; some features require billing. Auth is an API token from replicate.com/account/api-tokens (not created). Official Python client: pip install replicate and REPLICATE_API_TOKEN. HTTP API and OpenAPI schema are documented. Official MCP server and agent skills are documented at /docs/reference/mcp and /docs/reference/skills (pages were HTTP 200; packages were not installed). llms.txt is at https://replicate.com/llms.txt. A general OpenAI-compatible chat/completions guide was not found (paths such as /docs/guides/openai-api 404); do not assume drop-in OpenAI compatibility. Running models can generate media or otherwise spend credit. license proprietary; pricing paid (limited free use until billing is required). No signup or spend was used.
Agent access
Related
- Together AI: Hosted platform to run, train and serve open-source models via prepaid serverless, batch, dedicated and fine-tuning APIs, with an OpenAI-compatible base URL.
- Fireworks AI: Hosted inference and training platform for open models: prepaid pay-per-token serverless, reserved throughput, dedicated GPUs and fine-tuning, with OpenAI- and Anthropic-compatible APIs.
- fal: Generative media API plus serverless and dedicated GPU compute. Usage-based Model APIs and GPU-hour pricing; official key-based MCP.
Guides & comparisons
- GPU clouds and serverless inference for agents (Comparison): Compare GPU rental, serverless GPUs and hosted per-token APIs for agents: OpenAI-compatible base URLs, tools and structured output, BYO models, scale-to-zero and billing, APIs/CLIs, and host-trust notes. No cached dollar figures or invented benchmarks.
Sources
- https://replicate.com
- https://replicate.com/pricing
- https://replicate.com/docs
- https://replicate.com/docs/topics/billing
- https://replicate.com/docs/topics/billing/prepaid-credit
- https://replicate.com/docs/get-started/python
- https://replicate.com/docs/reference/http
- https://replicate.com/docs/reference/mcp
- https://replicate.com/docs/reference/skills
- https://replicate.com/docs/topics/security/api-tokens
- https://replicate.com/llms.txt
- https://replicate.com/terms
- https://replicate.com/privacy
Machine-readable
- JSON
- Markdown
- Source file on GitHub (edit via pull request)