AI coding assistants , by Anthropic , since 2025
Claude Code
Anthropic's terminal-based coding agent bundled into Claude Pro/Max subscriptions
- DESKTOP
- API
What Claude Code does
Claude Code is Anthropic's command-line and IDE agent that reads, edits, tests, and commits code across an entire repository with minimal hand-holding, powered by Claude Sonnet/Opus models. It ships bundled inside Claude's consumer Free/Pro/Max chat subscriptions, or can be billed separately through the Anthropic API.
What you get: Repo-wide autonomous editing, Terminal-native agent, Git-aware commits, IDE extensions, MCP support.
Typical jobs: Large refactors; Autonomous bug fixing; Codebase Q&A.
Worth knowing: Pricing page keeps exact Max-20x dollar figure and precise usage quotas undisclosed; only relative multipliers are published.
What users complain about: 'Anthropic tightens usage limits for Claude Code without telling users' (July 2025) (TechCrunch)
Pricing and plan limits
Verified August 2026| Plan | Monthly | Annual (per mo) | Limits |
|---|---|---|---|
| Free | Free | n/a | Claude Code included; capped low usage allowance, exact token/message quota not published |
| Pro | $20 | $204 | $17/mo effective on annual billing, $20/mo monthly; higher usage than Free on a 5-hour rolling window, exact quota not published |
| Max 5x | $100 | n/a | 5x Pro usage allowance |
| Max 20x | not disclosed | n/a | 20x Pro usage allowance; exact price not disclosed on the pricing page ("Starting at $100/mo" covers Max tier, 20x variant priced higher, not itemized) |
Where it wins
- Strong repo-wide agentic reasoning (Claude Sonnet/Opus) with native terminal and IDE integration
- Usage is bundled into an existing Claude Pro/Max consumer subscription rather than requiring a separate tool purchase
Where it falls short
- Anthropic tightened weekly and 5-hour usage limits multiple times through 2025-2026, in one case without prior notice to subscribers
- Heavy users report the effective cost of agentic sessions is far above self-hosted open-model alternatives (one developer cited roughly $100/month on Claude Code versus about $3/month running an open model backend)
How we scored it
Full methodology| Criterion | Weight | Score | Contribution |
|---|---|---|---|
| Capability | 30% | 10.0 | 3.00 |
| Value for money | 25% | 6.0 | 1.50 |
| Ease of use | 15% | 7.0 | 1.05 |
| Maturity | 15% | 8.0 | 1.20 |
| Support & docs | 10% | 8.0 | 0.80 |
| Momentum | 5% | 10.0 | 0.50 |
| Weighted total | 100% | 8.1 | 8.10 |
Alternatives to Claude Code
Chosen by editors, not sponsors-
LovableFreemiumClosest match on capability
-
AiderFreeCheaper entry plan
-
ClineFreemiumCheaper entry plan at $9.99/mo
-
GitHub CopilotFreemiumCheaper entry plan at $10/mo
Similar tools in AI coding assistants
AI-first code editor built on VS Code with agent mode and multi-file edits
- DESKTOP
AWS's IDE assistant with autocomplete, an agent, and automated Java upgrades
- PLUGIN
- DESKTOP
- API
- WEB
Browser AI agent that writes, provisions, and deploys full apps from a prompt
- WEB
- API