Guides → Playground & Guide → Multimodal RAG Playbook - Architecture, Cost, and Rollout for Vision/Audio/Text
Meet Theo Romero. Staff Engineer building a video + document Q&A product. "Three modalities (video, audio, text). Three architecture options. How do we pick - and how do we roll out without burning the budget?"
🔥 First prototype was $30K/mo at 100 users. Need to understand the architecture economics before scaling.
Multimodal RAG isn't 3 RAG pipelines bolted together. Vision, audio, and text each have different cost models, different failure modes, and different operational profiles. Picking the wrong architecture upfront costs 3-5× more to fix later.
Theo's prototype hit $30K/mo at 100 users - projecting $300K/mo at 1K users. Root cause: native multimodal embeddings called per-query for every image, no caching, no preprocessing. The playbook codifies which architecture to pick when, with cost ranges for each.
Three architecture patterns, three sweet spots. (1) Convert-to-text - transcribe audio, OCR images, then text-only RAG. Cheapest, simplest, may lose visual context. (2) Native multimodal embeddings - CLIP, multimodal embedding models. Most capable, most expensive. (3) Hybrid - image embeddings + text embeddings + audio transcripts, joined at retrieval. Most complex, often optimal at scale.
Complete playbook for multimodal RAG: architecture choices (convert-to-text vs native vs hybrid), cost projections, eval strategy, and rollout timeline.
multimodal_stack
Below: live sliders. Move them to see numbers change in real time. * Output uses the generic compute model — for precise numbers use the full calculator below.
Each input shapes your cost. Move the slider — see the impact.
Open the full calculator — pick a model, enter your tokens, see per-call, daily, monthly, and annual cost.
🚀 Open the full calculator →Audio costs are per-minute, dominated by transcription. $0.003-0.006/min × volume × 30 days.
Vision costs are per-image, vary 50× by detail level. Classification: $0.001/image. OCR: $0.01-0.05/image. Pick the lowest detail level your eval allows.
Text RAG is the cheapest line per unit. Standard pipeline. Embedding + storage marginal.
LLM read with multimodal context dominates if not optimized. Vision tokens count toward LLM input - passing images to LLM costs 10-50× more than text. Cache aggressively.
Same calculator, three different team sizes. Click a tab to see how the numbers shift.
Image-heavy small app. Native multimodal LLM. Modest scale. ~$800/mo.
Healthy range: $500-1.2K/mo
Theo's product after switching to convert-to-text + caching. ~$12K/mo down from $30K.
Healthy range: $8-15K/mo
Consumer-scale multimodal. Cheap-tier LLM mandatory. Multi-vendor routing. Self-hosted vector DB.
Healthy range: $200-500K/mo
Cost isn't the only dimension. Click any constraint — see how recommendations change.
Convert-to-text is the cost-effective default. Move to native multimodal only if eval shows quality benefit on your workload. Hybrid only at scale.
Multimodal hallucinations are sneakier. Eval each modality independently.
Compliance is per-modality. Vision-API BAA may differ from text-API BAA. Voice biometrics has its own regs.
Audio + images are highest-PII modalities. Strip metadata, get explicit consent, retention policy.
Multimodal queries are slower. Streaming UI helps. Pre-process modalities in parallel where possible.
Vision and audio APIs vary widely across vendors. Multi-vendor multimodal is harder than multi-vendor text.
Multimodal MLOps is genuinely harder. Eval frameworks for vision-RAG and audio-RAG are still maturing.
Tradeoff analysis is where most AI projects go sideways. Talk to a CFO-grade AI cost analyst →
Pre-loaded scenarios for the most common applications. Click a tab to see realistic numbers — then the "Try this scenario" button to load it into the calculator above.
Skip vision (visual content not needed for Q&A). Transcribe + chunk + embed transcripts.
Healthy range: $15-25K/mo
Papers with figures. Vision for figures + text for body. Premium LLM for reasoning.
Healthy range: $5-10K/mo
Image-similarity product search. CLIP-style embeddings. Cheap-tier LLM only for query disambiguation.
Healthy range: $15-30K/mo
Medical images + clinical guidelines RAG. HIPAA + self-hosted vector DB.
Healthy range: $3-6K/mo (premium tier)
Honest limitations — every model is wrong; some are useful. Where this one falls short:
For these, use: Multimodal RAG Stack for cost detail. Vision Cost + Audio Cost for per-modality math.
Author: Subu Vdaygiri, Founder & CEO of CloudIntelligence.ai. 17 years Fortune 100 (Ingram Micro, Siemens). Wharton CTO program · Kellogg CPO program · 10× AWS+Azure certified.
Why this matters: pricing for major vendors has dropped 40-90% in the last 24 months. A budget set 12 months ago is probably wrong by 30%+.
View 3-year history for →
Last-verified date is the most recent successful daily snapshot
(aicost_pricing_snapshots) or, when no snapshot exists yet,
the latest successful crawler run (aicost_crawler_runs).
10 of 10
vendors are currently verified. Aggregator services (TokenCost, AI Pricing Guru, etc.)
are not listed.
Derived from industry conventions, not directly published by the vendor. Typical conventions: cached input = 10% of base (90% off), Batch API = 50% of base (50% off).
| Vendor / Model | Field | Why it’s inferred |
|---|---|---|
| Anthropic — Claude Sonnet 4.6 | cachedInput |
Derived at 10% of input rate — Anthropic publishes 90% cache-hit discount on this tier. |
| Anthropic — Claude Sonnet 4.5 | cachedInput |
Derived at 10% of input rate; same 90% cache-hit convention as Sonnet 4.6. |
| Anthropic — Claude Sonnet 4.5 | batchInput |
Derived at 50% of standard input — Anthropic documents uniform 50% Batch discount. |
| Anthropic — Claude Sonnet 4.5 | batchOutput |
Derived at 50% of standard output — Anthropic documents uniform 50% Batch discount. |
| Anthropic — Claude Haiku 4.5 | cachedInput |
Derived at 10% of input rate — Anthropic 90% cache-hit discount convention. |
| OpenAI — GPT-5.4 Mini | cachedInput |
Derived at 10% of input — OpenAI documents automatic 90% discount on cache hits across GPT-5.x tier. |
| OpenAI — GPT-5.4 Nano | cachedInput |
Derived at 10% of input — OpenAI 90% cache-hit convention. |
| OpenAI — GPT-5.4 Nano | batchInput |
Derived at 50% of input — OpenAI Batch API uniform 50% discount. |
| OpenAI — GPT-5.4 Nano | batchOutput |
Derived at 50% of output — OpenAI Batch API uniform 50% discount. |
| OpenAI — GPT-5.4 Pro | cachedInput |
Derived at 10% of input — OpenAI 90% cache-hit convention. |
| OpenAI — GPT-5.4 Pro | batchInput |
Derived at 50% of input — OpenAI Batch API uniform 50% discount. |
| OpenAI — GPT-5.4 Pro | batchOutput |
Derived at 50% of output — OpenAI Batch API uniform 50% discount. |
| OpenAI — GPT-5.2 | cachedInput |
Derived at 10% of input; no residency uplift. |
| OpenAI — GPT-5.2 | batchInput |
Derived at 50% of input. |
| OpenAI — GPT-5.2 | batchOutput |
Derived at 50% of output. |
| OpenAI — GPT-5 | cachedInput |
Derived at 10% of input. |
| OpenAI — GPT-5 | batchInput |
Derived at 50% of input. |
| OpenAI — GPT-5 | batchOutput |
Derived at 50% of output. |
| OpenAI — GPT-5.5 Pro | cachedInput |
Derived at 10% of input — OpenAI does not publish a cached rate for *-pro models; using the family convention. |
| OpenAI — GPT-5.5 Pro | batchInput |
Derived at 50% of input. |
| OpenAI — GPT-5.5 Pro | batchOutput |
Derived at 50% of output. |
| OpenAI — GPT-5.2 Pro | cachedInput |
Derived at 10% of input — pro-tier convention. |
| OpenAI — GPT-5.2 Pro | batchInput |
Derived at 50% of input. |
| OpenAI — GPT-5.2 Pro | batchOutput |
Derived at 50% of output. |
| OpenAI — GPT-5.1 | batchInput |
Derived at 50% of input. |
| OpenAI — GPT-5.1 | batchOutput |
Derived at 50% of output. |
| OpenAI — GPT-5 Pro | batchInput |
Derived at 50% of input. |
| OpenAI — GPT-5 Pro | batchOutput |
Derived at 50% of output. |
| OpenAI — GPT-5 Nano | cachedInput |
Derived at 10% of input. |
| OpenAI — GPT-5 Nano | batchInput |
Derived at 50% of input. |
| OpenAI — GPT-5 Nano | batchOutput |
Derived at 50% of output. |
| Google — Gemini 3 Flash | cachedInput |
Derived at 10% of input — Google caching discount convention ~90%. |
| Google — Gemini 3.1 Flash-Lite | cachedInput |
Derived at 10% of input — Google caching convention. |
| Google — Gemini 3.1 Flash-Lite | batchInput |
Derived at 50% of input — Google Batch API uniform 50% discount. |
| Google — Gemini 3.1 Flash-Lite | batchOutput |
Derived at 50% of output — Google Batch API uniform 50% discount. |
| Google — Gemini 2.5 Pro | cachedInput |
Derived at 10% of input. |
| Google — Gemini 2.5 Flash | cachedInput |
Derived at 10% of input. |
| Google — Gemini 2.5 Flash-Lite | cachedInput |
Derived at 10% of input — Google caching convention. |
| Google — Gemini 2.5 Flash-Lite | batchInput |
Derived at 50% of input — Google Batch API uniform 50% discount. |
| Google — Gemini 2.5 Flash-Lite | batchOutput |
Derived at 50% of output — Google Batch API uniform 50% discount. |
| Google — Gemini 2.0 Flash | cachedInput |
Derived at 25% of input per Google 2.0 family caching rates. |
| Google — Gemini 2.0 Flash | batchInput |
Derived at 50% of input — Google Batch API uniform 50% discount. |
| Google — Gemini 2.0 Flash | batchOutput |
Derived at 50% of output — Google Batch API uniform 50% discount. |
| Google — Gemini 2.0 Flash-Lite | cachedInput |
Derived at 10% of input — Google caching convention. |
| Google — Gemini 2.0 Flash-Lite | batchInput |
Derived at 50% of input — Google Batch API uniform 50% discount. |
| Google — Gemini 2.0 Flash-Lite | batchOutput |
Derived at 50% of output — Google Batch API uniform 50% discount. |
| xAI — Grok 4 (legacy) | cachedInput |
Extrapolated at 25% of base. |
Pricing is cross-verified against the
LiteLLM community registry
when available. Daily snapshots are kept in aicost_pricing_snapshots;
every change is logged to aicost_price_changelog with old & new
values for full audit trail. Read the full methodology →