Generating a Nano Banana 2 image on BotHub on October 17, 2026, cost 13.65 rubles. On the Syntx platform, running the same model consumes one and a half internal tokens, but you cannot acquire them separately: under the Pro tier, the service sells 680 tokens for $17.96 per month. You cannot compare these two figures head-on: the internal Syntx token and CAPS in BotHub are different units of entirely different scales. In one system, a single operation requires tens of thousands of CAPS, while the other deducts a couple of tokens.
Fixed subscriptions to AI services frequently mask excessive expenses compared to pay-per-use models. Which option makes more sense for business: paying a subscription fee for extended limits or purchasing non-expiring request packages? The entire economics boils down to the nature of corporate workloads, since Syntx charges for a calendar period, whereas BotHub allocates funds based on actually executed operations.
Tariff Architecture and the Hidden Entry Barrier
Syntx's monthly subscription covers platform access, including tools and language models within the LLM studio. Meanwhile, generation tokens are billed separately, and the Basic, Pro, Elite, and Ultra Elite tiers feature options entirely devoid of generation tokens. On BotHub, the subscription fee is absent entirely, and balances decrease strictly as requests are sent, including queries to text neural networks.
In Syntx, money for an initial purchase is refunded only if you have not yet started using the service, and logging into the account counts as the start of use.
Such contract rigidity completely blocks refunds after the very first authorization. To avoid losses, testing capabilities requires utilizing the welcome 5 tokens and 5 requests in Syntx or spending the complimentary 30,000 CAPS in BotHub. Furthermore, the Syntx user agreement mandates disabling auto-renewal at least 24 hours before the end of the paid billing period.
Unlimited Models Versus a Flexible Balance
The argument favoring a fixed subscription fee lies in the convenience of continuous intensive work: subscriptions eliminate the need to constantly monitor account balances for every individual task. Pro tier subscribers receive unlimited access to GPT 5 Nano, DeepSeek V3, Claude Haiku 4.5, and Gemini 3.5 Flash Lite. On the VIP tier, the roster expands with Grok 4.3 and Qwen 3.7 Plus, Elite adds GPT 5.6 Luna and Gemini 3.5 Flash, and Ultra Elite unlocks Claude Sonnet 5.0 and Gemini 3.1 Pro. The DeepSeek V3 model remains accessible without restrictions even on the basic Basic plan.
However, promotional volume promises require sober recalculation.
Under the "Year" mode, the Basic card displays 3,120 tokens and "up to 10,404 images," alongside a price of $7.57 "per month."
Switching the interface to monthly billing reveals 260 tokens and a ceiling of up to 867 images for $9.41. The annual card aggregates yearly tokens with a monthly price, while the 867-image benchmark stems from a minimum calculated cost of 0.3 tokens per operation. Adding generation tokens increases the bill by $0.95 per month on the Basic tier and by $1.90 on the Pro tier, confirming that the primary share of the subscription covers the infrastructure itself.
What doesn’t add up
The complex split between BotHub CAPS and proprietary internal tokens acts primarily as a smoke screen that prevents any direct price comparison between single tasks. Strict terms that completely block refunds right after the first login further enforce this aggressive monetization model. Behind the loud marketing lists of unlocked models like DeepSeek V3 sits a routine fee charged just for access to platform tools rather than actual completed work. Adding generative tokens increases the monthly bill by 1.90 dollars on the Pro plan, while the subscription itself demands 17.96 dollars each month.