Plan event · 23 Oct 2026
All plan events · 31 providers tracked
Syntheticceiling
What moved
500 requests per 5-hour window per $30 pack, 1 concurrent request per model, up to 5 packs → Weekly credit quota only — 2% of the limit regenerates every ~202 minutes (a full week from zero), no 5-hour or monthly caps, 1 usage reset/month; tokens draw at subscription-discounted rates and more can be bought at usage-based prices
Synthetic replaced the request-pack limits with a single weekly credit limit that regenerates incrementally. The rate-limits page states the plans “only use weekly limits”, with no additional five-hour or monthly limits, 2% of the weekly quota returning every ~202 minutes and no lockout after a burst; each plan also carries 1 usage reset/month. The pricing page publishes a per-model token allowance per plan together with its API-rate dollar value, so the calculator now carries that published equivalent as each plan’s per-model monthly budget (Standard: $251/mo DeepSeek V4.1 Flash, $204/mo GPT-OSS 120B) where it previously showed n/a — Req/$1 reads as a value comparison (confidence proxy).
Verified 2026-10-23 · source
Affected plans
More Synthetic events
- 2026-10-23 · Price — synthetic.new/pricing now lists three subscription plans instead of one: Go $9, Standard $
- 2026-10-11 · Models — A new Standard-models row joined the Subscription roster: zai-org/GLM-5.3, marked Beta wit
- 2026-09-23 · Models — The synthetic.new/pricing Standard-models table dropped the zai-org/GLM-5.2 row and gained
- 2026-09-02 · Models — Synthetic.new Subscription ($30/mo) swaps Qwen3.6-27B for Qwen3.8-27B at unchanged rates a