Skip to content
Sovyron

Cheapest PDF-Input Model APIs

As of Nov 22, 2026, the cheapest PDF-capable model API is GLM-5.3-Flash at $0.0062 per 1M input and $0.0185 per 1M output tokens on JoinGonka.

The table ranks the 20 cheapest of 197 qualifying models offered by at least 3 providers, by the cheapest offer's 3:1 input:output blended price. Models that accept PDF files.

Cheapest PDF-Input Model APIs, USD per 1M tokens, as of Nov 22, 2026
#Model
1GLM-5.3-Flash$0.0062JoinGonka$0.0185JoinGonka$0.00931.0M113
2DeepSeek V4.1 Flash 1017$0.01Refuseless$0.08Refuseless$0.02751.1M73
3GPT-5.6 Luna$0.0123xxapi$0.0738xxapi$0.02771.1M54
4Ministral 3B$0.04Azure$0.04Azure$0.04131k7
5Gemma 3 4B IT$0.04Kilo Gateway$0.05Kilo Gateway$0.0425131k6
6DeepSeek V4.1 Flash$0.03engy$0.12engy$0.05251.1M82
7Gemma 3 12B IT$0.04Kilo Gateway$0.1Novita AI$0.06131k8
8GPT-6 Luna$0.04Kunavo$0.2Kunavo$0.081.1M48
9Gemma 3 27B IT$0.064Kilo Gateway$0.128Kilo Gateway$0.08131k11
10Gemma 4 26B A4B IT$0.05Kilo Gateway$0.19Kilo Gateway$0.085262k23
11Claude Haiku 5.5$0.05IteraCompute$0.25IteraCompute$0.11.0M47
12Ministral 3 3B$0.1Amazon Bedrock$0.1Amazon Bedrock$0.1262k3
13GPT-5 Mini$0.04QiHang$0.29QiHang$0.1025400k34
14Nova Lite 1.0$0.0595NanoGPT$0.238NanoGPT$0.1041300k5
15Nova Lite$0.06Amazon Bedrock$0.24Amazon Bedrock$0.1051.0M5
16Qwen3.8 27B$0.03CedarLLM$0.32engy$0.11381.1M65
17Muse Spark 1.3 Contributor$0.1Bothub$0.2Bothub$0.1251.0M10
18Muse Spark 1.2 Contributor$0.1CommandCode$0.2CommandCode$0.1251.0M9
19GPT-5 Nano$0.0455Poe$0.3636Poe$0.125400k32
20GPT-4o mini$0.075Cloudflare AI Gateway$0.3Cloudflare AI Gateway$0.1312128k25

FAQ

What is the cheapest PDF-capable model API?

As of Nov 22, 2026, the cheapest PDF-capable model API is GLM-5.3-Flash at $0.0062 per 1M input and $0.0185 per 1M output tokens on JoinGonka.

Which PDF-capable model APIs are cheapest after GLM-5.3-Flash?

DeepSeek V4.1 Flash 1017 ($0.0275 blended per 1M tokens on Refuseless), GPT-5.6 Luna ($0.0277 blended per 1M tokens on xxapi), Ministral 3B ($0.04 blended per 1M tokens on Azure), as of Nov 22, 2026.

How is this list ranked?

By each model's cheapest metered offer, blended 3:1: (3 × input + output) ÷ 4 per 1M tokens. Free tiers and subscription plans are excluded, and a model needs at least 3 paid providers to qualify.

Other cheapest lists