Nimbus-2 · long-horizon reasoning is live

Generative intelligence for the AGI era

Nimbus-2 plans, codes, and ships. One API call streams production-grade answers, decisions, and code at 842 tokens per second.

Start Building Read the Docs

Built for builders

Everything you need to ship generative features — nothing you don't.

A 1M-token context window with chain-of-thought planning that stays on task across entire codebases.

Typed SDKs for Python, Go, and TypeScript, structured JSON outputs, and token-level streaming by default.

Constitutional guardrails, automatic PII redaction, and full audit logs on every single request.

From API key to production in three steps.

Create a workspace, grab a key, and pick the model that fits your latency budget.

POST JSON to /v1/chat with tools, schemas, and system instructions attached.

Token-level SSE streams land in your app with a sub-200ms time-to-first-token.

“Nimbus-2 reviewed our ingestion pipeline and caught two race conditions. It reads a repo like a senior engineer.”

Mira Kato
Staff Engineer, Vector Labs

“We replaced three vendors with one Nimbus endpoint. Streaming JSON cut our p95 latency nearly in half.”

Dario Osei
CTO, Loopstack

Start free. Pay for tokens, not seats.

$0 / month

Start free

$49 / month

Choose Scale

Custom

Talk to sales

What is Nimbus-2?

Nimbus-2 is our frontier generative model: a 1M-token, tool-using reasoner tuned for coding, analysis, and long-horizon planning tasks.

How does pricing work?

You pay per token, never per seat. The Developer tier includes 1M free tokens each month; Scale adds priority routing and an SLA.

Is my data used for training?

No. API inputs and outputs are never used to train our models, and a zero-retention mode is available on every plan.

Get your API key in 60 seconds. No credit card required.

Free tier includes 1M tokens · SOC 2 Type II · GDPR ready