Abacus.AI ChatLLM (Abacus AI Agent, Desktop and CLI)

Abacus.AI · 2 plans · checked 2026-11-04

Plan	Price / month	Usage (as published)	Models
Basic	$10 $7 for the first month (promotion), then $10/month. Abacus AI Agent, Desktop/CLI and CoWork limited to 3 tasks of limited complexity a month.	20,000 credits per month; 3 Abacus AI Agent tasks per month.	Claude Fable 5.1, Claude Opus 5.5, Claude Sonnet 5.5, GPT-6.1 Sol, GPT-6 Astra, Gemini 4 Argon, Grok 4.7, Qwen 3.8 Max, DeepSeek V4.1, Kimi K3.1 Pro, GLM 5.3
Pro 1.5x Basic credits	$20 $20/month total (Basic $10 + Pro $10). Unrestricted Abacus AI Agent, coding agent and CLI while credits last; includes the SuperComputer (1 credit per 5 minutes running).	30,000 credits per month (20K Basic + 10K Pro).	Claude Fable 5.1, Claude Opus 5.5, Claude Sonnet 5.5, GPT-6.1 Sol, GPT-6 Astra, Gemini 4 Argon, Grok 4.7, Qwen 3.8 Max, DeepSeek V4.1, Kimi K3.1 Pro, GLM 5.3

How limits work: Monthly credit pool (Basic 20,000, Pro 30,000) shared by chat, Abacus AI Agent, Desktop/CLI, image/video generation and the RouteLLM API. Credits are not tokens; Abacus.AI does not publish per-model credit rates (shown only after use). After 75% of the monthly credits, the last 25% is meant for the last week of the billing month. An Agent task typically uses 500-1,000 credits.
Over the limit: Extra credits can be bought (the FAQ says 10K credits cost $10); purchased credits roll over. RouteLLM, GPT-5.4 Mini, Gemini 3.5 Flash Lite, GLM 5.3, GLM 5.3 Flash, DeepSeek V4.1 Flash, DeepSeek V4 Flash, DeepSeek V4 Pro, Kimi K2.7 Code, Muse Spark 1.3 and MiMo V2.6 Pro stay usable in ChatLLM after the credits run out (not via the RouteLLM API, except the RouteLLM router itself).
Official tools: Abacus AI CLI (abacusai, terminal UI and headless), Abacus AI Desktop (code editor, chat, CoWork), Abacus AI Agent (web)
Use in your own agent: yes: Claude Code, Codex CLI, Cursor, any OpenAI-compatible client (via the RouteLLM API, billed in credits)
EU-hosted: no

Notes

Abacus.AI, Inc. (USA). Per the RouteLLM docs all open-weight models are hosted on US servers; proprietary models are billed at the provider's API price with no markup.
Basic is $7 for the first month as a promotion (with fewer credits, e.g. 14K), then $10/month.
The CLI and Desktop can also run on your own OpenAI/Anthropic/Google API key, your ChatGPT plan's Codex usage, or a local/third-party provider such as Ollama or OpenRouter; those turns use no Abacus credits.
Teams: $10 per additional user per month; Enterprise from $5,000.
The billing FAQ's usage section still names older models (GPT-5.2, Sonnet-4.5) for the '15M input tokens' example; the model lists here come from the pricing page and Desktop FAQ.