FinBladeAI Pricing Calculator

Revenue stream	SaaS FB app · FB LLM	VPC Managed by Client FinBlade Hybrid	Onprem Air-Gapped Hybrid
Base License	No	Yes	Yes
Per-user license	Yes	Yes	Yes
Infrastructure (K8s Cluster)	No	No	No
Infrastructure Support	N/A	N/A	Optional
LLM Tokens (metered)	Yes	No	No
Storage	Yes	No	No
Model Updates / yr	Included	Included	Yes
Professional Services (one-time)	No	Yes	Yes

LLM models

Total tokens / month —

Use cases

One slider per workload: the track shows blue (input) and green (output) on the same bar; drag the thumb to set input tokens per user per month — output grows by a use-case-specific input-to-output ratio (see each row). Models below update from your non-zero use cases. As you drag, each row shows a usage level (Off → Light → Moderate → Heavy → Power) with a plain-language translation — e.g. queries, runs, or coding sessions per user per workday — so you can sanity-check the chosen volume. With Choose LLMs manually: check one or more models. The model with the orange outline is active — the sum of all sliders (each row’s input + output from its ratio) is saved to that model only. Check another model to shift the active slot and use the sliders again; per-model input and output appear under Total tokens / month.

Or choose models directly

Choose LLMs manually. Check models to include. The orange card is active: move the use-case sliders to assign that workload (sum of all sliders) to it. Check another model, or click an already-checked card, to shift activity and assign its own totals. See input/output per model under Total tokens / month.

Gemma 4 31B Dense Google open model: chatbots & assistants, long-document Q&A / RAG, multilingual apps—strong general quality on modest GPU footprints. Active: 31B (dense, all active) · Context: 256K tokens Architecture: dense · multimodal (text + image) GPT-OSS-120b OpenAI open-weight line: agent-style workflows, tool use, and reasoning—popular for private / on-prem deployment and customization. Active: ~5.1B/token (~117B total MoE) · Context: 128K tokens Architecture: MoE · text-only Qwen3.6-Plus Qwen Cloud API model (qwen3.6-plus): reasoning, coding, and agents—1M context for large codebases and docs, function calling + built-in tools, and hybrid thinking mode (per Qwen Cloud text-generation model table). Active: not listed · Context: 1M tokens · Max output: 64k · Thinking budget: 80k Architecture: hybrid linear attention + sparse MoE (per Qwen 3.6 release) · multimodal flagship on Qwen Cloud DeepSeek V4 Pro DeepSeek flagship: strongest math, coding, and reasoning—choose for complex dev tasks, agent-on-codebase, and deep analysis. Active: ~37B/token (~1T total MoE) · Context: 1M tokens Architecture: MoE · multimodal DeepSeek V4 Flash DeepSeek fast variant: optimized for speed and cost on coding, data queries, and high-throughput scenarios. Active: ~32B/token (~1T total MoE) · Context: 1M tokens Architecture: MoE · multimodal Kimi K2.5 Moonshot flagship (Jan 2026): agents, tools, coding, and enterprise automation—with native vision & video for UI/visual workflows and richer multimodal automation. Active: ~32B/token (~1T total MoE) · Context: 256K tokens Architecture: MoE · multimodal (text, image, video) OCR Document capture, not generative chat—use for scanned PDFs, forms, and images where you need reliable text extraction. Active: n/a · Context: n/a Architecture: not an LLM (OCR service)

Select a region above. Use the use-case sliders to drive models and token splits, or set sliders to zero and choose models manually below (add OCR separately if needed).

Sign in

FinBladeAI Pricing Calculator

Region

Deployment model

Support

LLM models

Use cases

Bill of Quantities