| Claude | Frontier assistants | Long-form writing, careful analysis, and coding if you'll pay for headroom. | Usage limits are the #1 complaint. Users report weekly caps draining in days, especially with Claude Code | Yes | Free tier: yes. Pro $17/mo billed annually or $20/mo monthly. Max from $100/mo. (checked Oct 10, 2026) |
| ChatGPT | Frontier assistants | People who want one assistant for writing, research, images and light coding. | Free and Go may show ads. Plus ($20/mo) is the cheapest ad-free plan | Yes | Free tier: yes. Go $8/mo (US, may include ads), Plus $20/mo, Pro from $100/mo. (checked Oct 10, 2026) |
| Google Gemini | Frontier assistants | People who live in Google Workspace. | Usage limits are compute-based and opaque, refreshing every 5 hours with a weekly cap | Yes | Free tier: yes. AI Plus $4.99/mo, AI Pro $19.99/mo, Ultra from $99.99/mo. (checked Oct 10, 2026) |
| Grok | Frontier assistants | Heavy X users who want real-time context. | No published usage numbers on the pricing page. Limits are described, not counted | Yes | Free tier: yes. SuperGrok $30/mo, SuperGrok Plus $100/mo. (checked Oct 10, 2026) |
| DeepSeek | Free and open models | Cheap reasoning and coding help on non-sensitive work, or local use. | Its privacy policy says data is processed and stored in China | Yes | Free tier: yes (app). API pay-per-token; see official pricing table. (checked Oct 10, 2026) |
| Perplexity | Multi-model platforms | Quick research where you want sources, and you'll actually check them. | Pro limits are 'weekly limits (average use)', with no number published | Yes | Free tier: yes (3 Pro Searches/day). Pro $20/mo or $200/yr. Education Pro $10/mo. Max $200/mo. (checked Oct 10, 2026) |
| Poe | Multi-model platforms | People who switch between many models. | Compute-point costs vary by model, and frontier models drain points fast | Yes | Free tier: yes. $4.99/mo (10k points/day) or $19.99/mo (660k points/mo). (checked Oct 10, 2026) |