Continue.dev review: replacing Google Antigravity IDE on a budget

An evaluation of Continue.dev as a highly configurable, low-cost alternative for developers facing quota exhaustion and model removals within the Google Antigravity IDE ecosystem.

The answer up front

For developers stranded by the recent Google Antigravity IDE update, which removed Gemini 3 and tightened quota limits, Continue.dev is the best escape hatch. It is an open-source VS Code and JetBrains extension that allows you to bring your own API keys. If you need to keep costs near zero, skip expensive proprietary IDEs and pair Continue with DeepSeek-V3 or Google's own Gemini Flash API. Skip Continue only if you require a fully managed, zero-config environment and are willing to pay a flat monthly premium for it.

Methodology

This review evaluates Continue.dev version 0.8.0 as a direct replacement for Google Antigravity IDE, prompted by the user signal from Reddit author aspiring_Dev_Ind. Our assessment draws on the technical architecture of Continue's open-source repository, its configuration schema, and public API pricing for Gemini and DeepSeek as of May 2026. This v0 review relies on documented configuration capabilities and API cost structures. Independent latency and code-generation quality benchmarks under heavy local workloads are pending. We will update this review if API pricing structures or local model execution performance shifts significantly.

What it does

Continue.dev operates as an open-source side panel and inline edit assistant inside VS Code or JetBrains. Unlike Google Antigravity IDE, which locks users into a managed subscription and specific internal models, Continue acts as an orchestration layer.

Bring your own keys

The core design of Continue relies on a local config.json file. You input your own API keys for providers like Google AI Studio, Anthropic, or DeepSeek. This bypasses the hard quotas imposed by managed IDE plans. If you exhaust your quota on one model, you switch the active model in the dropdown instantly without changing your development environment.

Local and remote flexibility

Continue supports both remote APIs and local execution. You can run lightweight models locally via Ollama or Llama.cpp, which costs nothing. For more complex reasoning, you can route queries to commercial APIs. This hybrid setup directly addresses the budget constraints of developers who cannot afford flat-rate $20 monthly plans.

What's interesting and what's not

The most compelling aspect of Continue is its complete lack of vendor lock-in. When Google Antigravity IDE removed Gemini 3, users had no recourse but to accept the downgraded quota. With Continue, if a provider deprecates a model or raises prices, you edit a single line of JSON to point to a different LLM provider.

However, this flexibility comes with a configuration tax. There is no polished, one-click onboarding. You must manually generate API keys, manage billing accounts with cloud providers, and fine-tune system prompts in your local configuration. For developers accustomed to the turnkey experience of Google Antigravity IDE, this setup phase can feel tedious. Furthermore, Continue lacks the deep workspace indexing that some proprietary IDEs offer out of the box, requiring manual indexing configurations for large codebases.

Pricing

Continue.dev is entirely free and open-source under the Apache 2.0 license. Your only costs are the direct API usage fees from your chosen providers. Pricing snapshot date: May 2026.

• Continue.dev Extension: $0
• DeepSeek-V3 API (External): $0.14 per million input tokens, $0.28 per million output tokens
• Google Gemini 1.5 Flash API (External): $0.075 per million input tokens, $0.30 per million output tokens

Verdict

Continue.dev is the most practical alternative for developers fleeing the quota constraints of Google Antigravity IDE. By shifting from a flat-rate IDE subscription to a pay-as-you-go API model using DeepSeek-V3 or Gemini Flash, budget-conscious builders like aspiring_Dev_Ind can reduce their monthly spend to under $5 while maintaining high-quality code generation. It requires some initial configuration, but the long-term autonomy and cost savings make it the superior choice over locked-down proprietary platforms.

What we'd test next

In our next phase of testing, we plan to benchmark the latency of Continue.dev when processing large codebases of over 50,000 lines. We will specifically measure the context-window retrieval accuracy when using Gemini 1.5 Flash via API versus the local indexing performance of Continue's codebase embeddings.

The investor read

The migration of developers from closed, flat-rate IDEs like Google Antigravity to open-source orchestration layers like Continue.dev highlights a critical shift in developer tooling spend. As frontier model providers squeeze margins and restrict quotas on flat-rate plans, developers are realizing that 'bring-your-own-key' (BYOK) architectures offer superior unit economics. For investors, this signals that the value in the AI developer toolchain is bifurcating: value either accrues to the model providers directly via API consumption, or to highly integrated, proprietary platforms that offer proprietary context-indexing engines that open-source extensions cannot easily replicate. Pure-play wrapper IDEs charging a flat $20/month fee are highly vulnerable to margin compression and user churn when model access terms change.

Pull quote: “Unlike Google Antigravity IDE, which locks users into a managed subscription and specific internal models, Continue acts as an orchestration layer.”