Open-source VS Code and JetBrains extension connecting any LLM for chat, autocomplete, and agentic coding.
No compliance attestations on file. Confirm directly with the vendor before procurement.
| Tier | Price | Includes |
|---|---|---|
Open Source | Free | VS Code and JetBrains extensions, all features (chat, autocomplete, edit, agent), bring your own API keys with any supported model provider |
Starter | $3 per million tokens (input + output), pay as you go. Includes Continue Hub agents, integrations like Slack/Sentry/Snyk, frontier model credits. | — |
Team | $20/seat/mo | — |
Company | Contact sales | — |
What it does Continue is an open-source (Apache 2.0) AI coding extension for VS Code and JetBrains. Core features: autocomplete, chat, edit, and agent mode powered by any LLM you connect (Anthropic, OpenAI, Gemini, Bedrock, Azure, Mistral, Ollama, Bedrock, self-hosted). Continue Hub adds a registry of shareable AI assistants, agents, rules, prompts, and MCP servers.
Who it's for Developers and platform teams who want IDE AI assistance without vendor lock-in, particularly when they need to bring their own model (regulated environments, cost optimization with Ollama, on-prem inference). Also AI engineers building custom code assistants who want a configurable open-source baseline.
How platform engineers use it
Install the Continue extension in VS Code or JetBrains. Configure config.yaml with your model providers — Claude for chat, a fast local model for autocomplete via Ollama, Bedrock for compliance-bound usage. Wire MCP servers for tool use. For teams, use Continue Hub to share standardized agent configurations, custom rules, and prompts across the org. The Team plan adds private agent sharing and SSO; Company tier adds BYOK at the workspace level and SLA.
Strengths
Limitations
AI maturity AI-native. Continue was one of the first open-source LLM IDE extensions and has continuously expanded as model capabilities matured (autocomplete → chat → edit → agent → MCP). The roadmap into Continuous AI and Mission Control is real product expansion, not vapor. Treat as the open-source default that works with whatever models you're committed to.
