1 / 3
【Overview】 RunInfra is a hosted inference platform for open-weight chat models, built with AI agents in mind. It exposes an OpenAI-compatible chat completions endpoint and an Anthropic-compatible Messages endpoint under a single API key, so teams can point existing OpenAI or Anthropic SDK code at open models without rewriting their integration layer. The platform also adds session-aware routing and prompt caching aimed at the pattern where an agent resends the same conversation context on every tool call. The problem RunInfra targets is the cost and latency of repeated context. When an agent replays a growing conversation on each step, most of that input is unchanged from the previous call. RunInfra routes repeat calls within a session to the replica that already holds the cached context, so the repeated portion bills at the cached input rate rather than the full input rate. Streaming, tool calling, and JSON mode are supported on chat model requests, and some models accept image input. RunInfra is aimed at AI agent builders, startups from bootstrapped through Series A (which can apply to RunInfra for Startups), and developers who work inside coding agents such as Claude Code and OpenCode. It is not intended for existing RunInfra customers seeking the startup program, or for workloads that process protected health information. 【Key features】 • Hosted open-model inference: RunInfra serves open-weight chat models through its Model APIs, so there is no GPU fleet to operate. • OpenAI and Anthropic SDK compatibility: The same API key works with OpenAI-compatible chat completions and Anthropic-compatible Messages. • Session-aware routing and prompt caching: Repeat calls in an agent session land on the replica holding the cached context, so repeated input bills at the cached input rate. • Coding plans: Monthly plans include usage for Claude Code, opencode, and 20+ agents, with limits that reset every 5 hours and weekly. • Streaming, tool calling, and JSON mode: Chat requests can stream responses, call tools, and return structured output. • Image input: Some models accept images alongside text. 【Use cases】 1. Running AI agents that replay conversation context on every tool call. 2. Setting up open models inside a coding agent. 3. Calling hosted open-weight chat models through the OpenAI or Anthropic SDKs. 【Pricing】 Pay as you go: $5 — any amount from $5, pay per token at published prices, credits last a year, with OpenAI-compatible chat completions and Anthropic-compatible Messages. Starter: $10/month — about $12 of usage a month, 5-hour limit $1.11, weekly limit $2.77, free Standby up to $0.20 of chat a day. Pro: $29/month — about $35 of usage a month, 5-hour limit $3.23, weekly limit $8.08, free Standby up to $0.58 of chat a day. Team: $99/month — about $119 of usage a month, 5-hour limit $10.98, weekly limit $27.46, free Standby up to $1.98 of chat a day. Enterprise: Custom — dedicated infrastructure, compliance, and custom volume, audit logs and role-based access control (RBAC), custom volume and contract terms, and custom SLAs up to 99.99%. 【Pros and cons】 Pros: - Agent builders whose workloads resend context on every tool call get the most value, since cached input pricing applies directly to the repeated portion. - Teams already using the OpenAI or Anthropic SDK can try open models with minimal code change, which lowers the cost of evaluating a switch. - Coding-plan users get included usage with short reset windows, which suits bursts of agent-driven coding rather than steady round-the-clock load. Cons: - Cache hits are best effort, so cached-input savings are not a guaranteed outcome on every call. - Coding plans cover chat only; images and audio always stay pay as you go. - JSON mode depends on the individual chat model rather than being available uniformly across the catalog. 【FAQ】 Q: How much does RunInfra cost to start? A: Pay as you go starts at any amount from $5 in credits, which last a year and bill per token at published prices. Monthly plans run $10 for Starter, $29 for Pro, and $99 for Team, with Enterprise pricing custom. Q: Who is RunInfra built for? A: It is aimed at AI agent builders, startups from bootstrapped through Series A, and developers using coding agents like Claude Code and OpenCode, especially where agent traffic resends context on every tool call. Q: Do coding plans include everything I can send? A: No. Coding plan coverage is chat only; images and audio are always billed pay as you go. Cache hits under session-aware routing are also best effort.
Pay as you go · $5 (Any amount from $5, Pay per token at published prices, Credits last a year) Starter · $10/month (About $12 of usage a month, 5-hour limit $1.11, Weekly limit $2.77) Pro · $29/month (About $35 of usage a month, 5-hour limit $3.23, Weekly limit $8.08) Team · $99/month (About $119 of usage a month, 5-hour limit $10.98, Weekly limit $27.46) Enterprise · Custom (Dedicated infrastructure, compliance, and custom volume, Audit logs and role-based access control (RBAC), Custom volume
WPS AI is an AI assistant built into WPS Office that helps users write documents, generate presentation outlines and…
Baidu Comate is an AI coding assistant built by Baidu on its ERNIE model. It goes beyond simple autocomplete: it…
UniConverter is an AI media conversion and enhancement tool for video, audio, and images. It combines everyday…
6 similar tools to consider instead of RunInfra