Free for 14 days, then a flat monthly price. No metering, no per-token invoice, no card until you actually run out of the trial.
| Model | Price | Context | Good at |
|---|---|---|---|
GLM 5.3 glm-5.3 |
$4 / month | 1M | agentic work with tools, refactoring, scripts |
DeepSeek V4 Pro deepseek-v4-pro |
$5 / month | 1M | debugging, math and formal tasks, surgical edits |
KIMI K3.1 Pro kimi-k3.1-pro |
$6 / month | 1M | whole repositories, long-form writing, voice and role |
One key works on all three; the price is for the model you keep switched on. Nothing is metered — a long agent day costs the same as a short one. That is the whole point of not billing by token.
| Claude Pro — $20 / month | AbliterateMe | |
|---|---|---|
| Price | $20 / month | $4–6 / month |
| Requests per 5-hour window | ~45 | ~450 |
| Agents in parallel | 1 | 10 |
| Context window | 200K | 1M |
| Third-party endpoint | not allowed | drop-in |
10× Claude Pro plan usage at a fifth of its price, because there is no frontier lab to amortise: the weights are open, we run them ourselves, and we are not paying for the next release. What you give up is raw capability on the hardest problems — see what runs here.
| Trial | Standard | Team | |
|---|---|---|---|
| Price | $0 — 14 days | $4–6 / mo | from $25 / mo |
| Models | all three | all three | all three |
| Usage limits | 10× Claude Pro plan usage | 10× Claude Pro plan usage + priority | Reserved |
| Concurrency | Shared, rate limited | Priority | Reserved |
| Support | — | Email, 1 day |
If usage gets absurd we will rate limit you rather than bill you. The trial never converts by itself and we never take a card without asking.