AI Cost Optimizer ✦ New

Cut your AI bill by up to 80%

A drop-in proxy that optimizes images, audio, and video before forwarding to OpenAI or Anthropic. Fewer tokens billed, same answer quality. You pay normal provider rates on the optimized file — we charge 20% of savings.

See pricing →Talk to us

Inspect endpoint always free · $10 credit on sign-up · No monthly fee on pay-as-you-go.

92%

Saved on product images

Task-aware compression pipeline

35%

Saved on audio transcription

Intelligent silence & noise filtering

75%

Saved on video understanding

Quality-weighted frame selection

20%

MediaLayer fee

% of savings only — zero upfront

Try it now

See your savings in seconds

Drop in any image or audio file. The inspect endpoint runs the full optimizer pipeline and shows exactly how many tokens would be saved — without making a real AI call.

Inspect playground

Dry-run — no AI call, no account needed

Always free

Drop image here or click to browse

JPEG, PNG, WEBP, BMP — max 1 MB

Prompt

Provider

5 inspections / hour · No account required · Sign up to get $10 free credit for live processing

How it works

One API call. Automatic savings.

Replace your direct OpenAI / Anthropic call with a MediaLayer call. Same prompt, same response shape — just cheaper.

Send your media

POST your image, audio, or video to the MediaLayer endpoint with your prompt and provider choice (OpenAI or Anthropic).

Optimizer runs

The pipeline detects the task, then applies the right compression, resize, silence removal, or frame selection — in 10–50 ms.

Forwarded to the AI provider

The optimized file and your prompt are sent to OpenAI or Anthropic. The full JSON response comes back to you unchanged.

You only pay for optimized tokens

Your provider bill reflects the smaller file. MediaLayer invoices 20% of the dollar savings at end of month.

Media types

Optimize every media type

Image

40–92% token savings

Task detection — OCR, describe, classify, extract
Precision resize to provider tile boundaries
Perceptual quality tuning without visible loss
Smart detail-mode selection per task type

JPEG, PNG, WEBP, BMP, TIFF, GIF

Audio

20–45% token savings

Intelligent silence & low-energy segment removal
Adaptive resampling optimized per provider model
Overlapping chunk analysis for accuracy
Language-aware preprocessing

MP3, WAV, FLAC, OGG, AAC, M4A

Video

60–80% token savings

Quality-scored frame selection — only informative frames
Scene-change detection & static frame elimination
Temporal coverage scoring across the full clip
Per-frame compression after scene analysis

MP4, MOV, AVI, MKV, WEBM

API

One endpoint. Drop-in replacement.

Swap your direct provider call for a MediaLayer call. Same prompt, same response structure — but the file sent to the provider is already optimized.

Multipart form data — image, audio, or video file
Returns full provider JSON response unchanged
Optimization report included in every response
`/inspect` endpoint for dry-run savings preview (free)
X-API-Key auth — no provider keys in your requests

See pricing →Request a demo

POST /v1/ai/process — response

{
  "result": "A product photograph of white sneakers...",
  "optimization": {
    "task_detected": "describe",
    "detail_mode": "low",
    "optimization_tier": "aggressive",
    "original_tokens": 1105,
    "optimized_tokens": 85,
    "original_cost_usd": 0.003315,
    "optimized_cost_usd": 0.000255,
    "savings_pct": 92.3
  },
  "timing": {
    "preprocess_ms": 18,
    "provider_ms": 412,
    "total_ms": 430
  }
}

Pricing

Pay only when you save

The inspect endpoint is always free. Process calls are priced as a share of the savings we generate — no savings, no fee.

Always free

Inspect

$0forever

See exactly what the optimizer would do to your media and how much it would save — without making a real AI call.

Unlimited dry-run requests
Full savings estimate per file
Task detection + optimization report
Image, audio, and video
No API key needed for inspect

Get started free

Frequently asked questions

How does the 20% savings fee work?

At the end of each month we tally the total dollar savings generated across all your process calls — the difference between what you would have paid the AI provider without optimization and what you actually paid. We invoice 20% of that number. If we save you $100 in OpenAI costs, you owe us $20.

What if the optimizer doesn't save anything on a particular file?

No savings means no fee. We only charge when we reduce your token bill. You always pay less than you would without the optimizer.

Do I still get billed by OpenAI or Anthropic directly?

Yes. Your provider API key is stored encrypted on our servers. OpenAI or Anthropic bills you directly for the optimized token usage. MediaLayer charges a separate monthly invoice for 20% of the savings we created.

Is the inspect endpoint really free?

Yes, completely free and unlimited. The inspect endpoint runs the full optimization pipeline and returns a savings estimate without making a real AI provider call. Use it to evaluate how much you'd save before committing to process calls.

Does optimization affect answer quality?

For most tasks, no. The optimizer is task-aware — it uses different strategies for OCR (preserves resolution), describe (can compress more), count (keeps detail mode), etc. The optimization parameters are tuned per task type to preserve the answer quality you would have gotten from the unoptimized file.

Start cutting your AI costs today

The inspect endpoint is free forever. Sign up and get $10 in free processing credit — no credit card required to start.

Get $10 free credit →Try inspect playground

No credit card · No monthly fee · Pay 20% of savings only