> 📥 The full one-pager is available as a PDF (Danish). Download it at the top of the page — four tools side-by-side, plus three questions before you decide.
Which coding agent do you run in your terminal?
Choosing the tool and choosing the model are two different decisions. The tool defines your workflow (terminal, IDE, multi-agent, procurement). The model defines capability and where data is processed. And as Google just reminded us with Gemini CLI: the vendor can change the rules on 30 days' notice. Here's the status of the four dominant tools as of June 22, 2026.
01 · Claude Code (Anthropic)
Today's flagship: claude-opus-4-8. Terminal-first, built for long autonomous sessions with full context and deep refactoring.
Strengths
- Best-in-class on long, coherent coding tasks and large refactors.
- Available via AWS Bedrock and Google Vertex — EU region is an option.
- Skills, hooks, subagents; a mature CLI.
Watch out
- Fable 5 (Mythos class, 1M context) launched June 9 but has been globally suspended since June 12 following a US export-control directive. The real flagship right now: Opus 4.8.
- No free tier; programmatic use draws on separate credits.
- 🚩 The export-control case affects foreign nationals (including in the EU) — frontier access can shift day to day. A vendor risk, not just a feature difference.
02 · Codex (OpenAI)
Model: gpt-5.5. Bundled with the ChatGPT subscription and covers the widest surface — from terminal to GitHub.
Strengths
- Widest surface coverage: CLI, IDE, ChatGPT cloud, GitHub bot, and computer use — one account ties it together.
- Token-efficient on GPT-5.5; included in Plus/Pro/Business/Enterprise.
- ~4 million weekly developers — large ecosystem.
Watch out
- Best if you're already in the ChatGPT ecosystem; otherwise API prices on top.
- Usage limits/credits can bite on large, long-running tasks.
- 🚩 GPT-5.5 is strong on cyber. Clarify your policy on offensive vs. defensive use cases before broad rollout.
03 · Antigravity (Google)
Model-agnostic: Gemini 3.x · Claude 4.6 · GPT-OSS. The most ambitious on multi-agent orchestration. It doesn't only run Gemini — you pick the model.
Strengths
- Multiple agents in parallel (Manager view); free model choice per task in one environment.
- Artifacts make agent work traceable; CLI + SDK + desktop share the same harness.
- Free preview with generous Gemini limits.
Watch out
- New and changing fast; free tier heavily cut (~20 requests/day, weekly cap).
- Replaces Gemini CLI, which shut down on June 18 — forced migration and community pushback (closed source).
- 🚩 Google killed an open-source tool with 100k+ stars on 30 days' notice. Example of how fast the rules can change.
04 · GitHub Copilot (Microsoft)
Multi-model: GPT · Claude · Gemini. Often wins on procurement rather than raw capability. Runs several model families under the hood.
Strengths
- Strongest on compliance & procurement; deeply integrated in the GitHub flow.
- Now US + EU data residency (Microsoft's EU Data Boundary, incl. EFTA/Nordics) — also for Claude models. FedRAMP for the public sector.
Watch out
- Lower capability ceiling than the pure frontier tools.
- EU residency is opt-in (off by default, ~10% higher AIC usage); the newest models can lag in regional availability. Residency ≠ full data-processor compliance.
- 🚩 "Claude in Copilot" is now blue-stamp-able in the EU — but only when your admin actively turns the residency policy on.
When to pick what
| Tool | Pick it when … |
|---|---|
| Claude Code | long autonomous sessions and deep refactoring in the terminal are the core — and you can live with model uncertainty at the top right now. |
| Codex | you already pay for ChatGPT and want one tool from terminal to GitHub. |
| Antigravity | you want to orchestrate multiple agents and pick the model per task. |
| GitHub Copilot | procurement, compliance, and EU data residency weigh more than the absolute capability ceiling. |
Three questions before you pick
1. Can you switch vendor in a week — or are you locked into a quarter-long migration?
2. Where is data processed during inference, and does that meet your residency requirements?
3. Who owns model selection — IT or the individual developer?
Sources
Google Developers Blog (Gemini CLI → Antigravity, May 19) · Anthropic (Fable 5/Mythos 5 + suspension, June 9–12) · OpenAI (GPT-5.5 / Codex) · GitHub Changelog (data residency, April 13).
The field moves weekly — check primary sources before any decision.
> ⚠️ Use at your own responsibility. This one-pager is a professional snapshot as of June 22, 2026 for informational and educational purposes and does not constitute legal, regulatory, or procurement advice. Prices, models, and vendor terms can change without notice — always verify primary sources before a decision.
Vendor-neutral · human-in-the-loop by default · EU AI Act-aware.
📬 Subscribe to the newsletter "AI, Built Human" on Substack — this field changes every week.
Stefano Vincenti · AI Advisor & External Lecturer, IT University of Copenhagen · Cofounder & CTO BotTellMe · Partner, TryZone · aitrainer.dk