API and service
OpenAI: latest release, open vulnerabilities and status today
The latest release of the openai package is 7.36.0. No advisories for the openai package were published in the last 12 months. OpenAI Status shows no open incidents right now.
Last checked against npm, OSV.dev and OpenAI Status.
- Latest openai release
- 7.36.0
- npm downloads last week
- 43.4M
- Licence
- Apache-2.0
Security advisories
No advisories published in the last 12 months. Checked packages: openai.
Status now
No open incidents or maintenance right now.
11 incidents in the last 90 days (2 major). Most recent major: “Widespread feature degradation” (2026-10-15). Not counted: 1 notice marked no impact.
Production checklist
OpenAI production checklist: 5 checks before launch
- Must do: Call the API from your server only; never ship the key to clients
- Should do: Set project budgets and per-user rate limits
- Should do: Treat model output as untrusted input
Alternatives
Anthropic Claude
Advanced large language model with extended context and reasoning
Strengths
- 200K+ token context window
- Strong reasoning and analysis capabilities
Trade-offs
- Smaller fine-tuning options
- Less extensive plugin ecosystem than GPT
Google Gemini
Google's multimodal AI model family
Strengths
- Native multimodal (text, image, video, audio)
- Long context window (1M+ tokens in Pro)
Trade-offs
- Less established than GPT/Claude
- Google ecosystem dependency
Cohere
Enterprise-focused LLM platform
Strengths
- Enterprise-grade security and compliance
- Strong at text generation and summarization
Trade-offs
- Less well-known than OpenAI/Anthropic
- Smaller model ecosystem
Meta Llama
Open-source large language models
Strengths
- Fully open-source (can self-host)
- Strong community and fine-tunes
Trade-offs
- Requires infrastructure for hosting
- Lower performance than closed-source models
Example stacks with OpenAI
Docs and release notes
Check OpenAI in your own stack
This page checks OpenAI in general. Your repo pins specific versions: the free dependency check reads your manifests and lists what to fix first. Prism Agent runs the same live checks on your whole stack and turns them into a plan your coding agent can build.