ToolsCompaniesAbout

CoreWeave Serverless Inference

by CoreWeave@coreweaveNo ratings yet

Pay-per-token API access to open-source models and LoRA-tuned weights, with built-in tracing and evals.

Visit coreweave.com
Tool type
Pay-per-token inference API for open-source and LoRA-tuned models
Made for
AI engineers, AI application teams, and ML platform teams
Runs on
CoreWeave-hosted serverless API, accessed via web playground and API endpoint
Requests are served by CoreWeave; no cluster or GPU selection is required.
Last checked
29 Oct 2026
2 product sources

What it does

CoreWeave Serverless Inference, part of the Forge platform, takes API requests carrying prompts or chat messages (including image inputs for multimodal models) against a curated catalog of open-source models, or against your own LoRA weights served alongside them. It returns model completions, and each request can be traced and evaluated through native observability features without extra instrumentation.

People use it to test and swap models quickly without provisioning clusters or choosing GPUs. The catalog includes models such as DeepSeek, GLM, Kimi, Qwen, Gemma, Llama, and GPT OSS variants, and the playground lets you compare them before writing code. Requests go to an OpenAI-compatible endpoint, so an existing SDK can be used, and billing is pay-per-token with a spend dashboard and budget controls.

It suits AI engineers building copilots, agents or retrieval applications, and teams serving supported LoRA adapters. Access starts with a Forge account and the inference documentation. The service manages model serving, while the application supplies requests and handles the returned results. Charges depend on the selected model and input, cached and output tokens. Check catalog availability, rate limits and budget settings before sending production traffic.

What it can do

  • Serve open-source models from a curated catalog through a single API
  • Serve your own LoRA weights alongside catalog models without changing your integration
  • Compare and test models in a playground with no endpoint configuration or keys
  • Trace every request and run evals and production monitoring without extra instrumentation
  • Use an OpenAI-compatible endpoint with an existing SDK
  • Support tool calling and structured outputs on supported models

Works with

  • OpenAI-compatible API endpoint (existing SDKs)

Example workflows

Swap models during app iteration
Illustrative example. Input: a prompt set and two candidate catalog models. Steps: send the same requests to each model's endpoint via the OpenAI-compatible API, then review the traced requests and evals. Output: a side-by-side view of responses, latency, and token usage for choosing a model.
Serve a fine-tuned LoRA model
Illustrative example. Input: your LoRA weights for a supported base model. Steps: attach the weights to the serverless endpoint, send requests with the existing SDK, and monitor traced production requests. Output: completions from your fine-tuned model served without building a deployment pipeline.

Before you choose

  • Pricing is per model and per token (input, cached, and output rates differ), so check the model's rate before committing to a workload.
  • Catalog-only serving: custom weights are supported via LoRA; full custom-model deployment is a different product (Dedicated Inference).
  • Check current model availability, rate limits and concurrency allowances before planning production traffic.

Listed under

Get started

Start with the getting-started guide and the model catalog.

  1. Create a CoreWeave account at the sign-up page.
  2. Open the getting-started guide and choose a model from the catalog or the playground.
  3. Send a request to the OpenAI-compatible endpoint with your existing SDK and review traces in the observability view.
Start with Serverless Inference

Plans and prices

Prices in USD

Inference is charged by model and token usage. Input, cached-input and output rates can differ. Forge software plans and temporary credits do not make inference usage free; consult the current model rate table.

Pay-per-token inferenceUsage-basedRates vary by model; see the model rate table.
  • Curated open-source model catalog
  • OpenAI-compatible API
  • Built-in tracing and evals

Source: coreweave.com/products/coreweave-forge/serverless-inference, checked 29 Oct 2026. Prices can change; check before you buy.

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

Ratings and reviews

No ratings for CoreWeave Serverless Inference yet.

Be the first to rate it

What’s changed

Free account

See what’s changed at CoreWeave Serverless Inference

Launches, price moves and pivots, each with a date and a source. Free with an account.

Signals

Free account

See CoreWeave Serverless Inference’s signals

Launch traction, hiring and release activity, tracked over time. Free with an account.

Data and security

Free account

See how CoreWeave Serverless Inference handles your data

Training on your files, compliance, hosting and the models behind it. Free with an account.

Compare with similar tools

Free account

Compare CoreWeave Serverless Inference side by side

Prices, platforms and features against similar tools, with sources. Free with an account.

Where this comes from

How we research
Company website
coreweave.com: product, pricing and policy pages
YC directory mirror
Batch and company name
Social accounts
Accounts the company links, and its posts on X
Human review
Checked by a person
knowledge_base

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.