Ignite Models · Rate Sheet

Per-model rate sheet.

Every model in the catalog with its billing mode, multiplier, and effective price. Token-billed models multiply the base credit rate per model; ONNX models bill at Ignite Apps compute rates.

Base rates · how effective price is computed

LLM prompt
$0.05 / 1M credits
LLM completion
$0.40 / 1M credits
Embeddings
$0.03 / 1M tokens
ONNX / CPU-time
$0.0432 / vCPU-hr

Effective = tokens × cost_multiplier × base. Multipliers live in each model's model.yaml in ignite-catalog/models/registry.

Chat & reasoning · 3

Model	ID	Mode / Multiplier
Kimi K2.5 kimi-k2.5 · openai-compatible	15.6× · 8.125×	$0.78 / 1M prompt $3.25 / 1M completion
Kimi K2 kimi-k2 · openai-compatible	15.6× · 8.125×	$0.78 / 1M prompt $3.25 / 1M completion
Moonshot V1 Auto moonshot-v1-auto · openai-compatible	15.6× · 8.125×	$0.78 / 1M prompt $3.25 / 1M completion

Embeddings · 4

Model	ID	Mode / Multiplier
Jina Embeddings v4 jina-embeddings-v4 · runpod-gpu	1×	$0.03 / 1M embedding tokens
Snowflake Arctic Embed M v2.0 arctic-embed-m-v2 · onnx-embedding	cpu_time	Ignite compute rate$0.0432 / vCPU-hr + $0.0060 / GB-RAM-hr
Google EmbeddingGemma 300M embeddinggemma-300m · candle-embeddings	cpu_time	Ignite compute rate$0.0432 / vCPU-hr + $0.0060 / GB-RAM-hr
ArcFace ResNet-100 (face) arcface-r100 · onnx-embedding	cpu_time	Ignite compute rate$0.0432 / vCPU-hr + $0.0060 / GB-RAM-hr

Reranking · 1

Model	ID	Mode / Multiplier	Effective rate
Jina Reranker v2 jina-reranker-v2 · runpod-gpu	1×	$0.05 / 1M prompt

Audio transcription · 2

Model	ID	Mode / Multiplier	Effective rate
Whisper Large v3 Turbo (GPU) whisper-large-v3-turbo · runpod-gpu	1×	$0.05 / 1M prompt $0.40 / 1M completion
Whisper Large v3 Turbo (CPU) whisper-large-v3-turbo-cpu · candle-transcription	1×	$0.05 / 1M prompt $0.40 / 1M completion

Translation · 1

Model	ID	Mode / Multiplier	Effective rate
Seed-X PPO 7B Translation seed-x-ppo-7b · runpod-gpu	1×	$0.05 / 1M prompt $0.40 / 1M completion

OCR · 2

Model	ID	Mode / Multiplier	Effective rate
PaddleOCR-VL (GPU) paddleocr-vl-gpu · runpod-gpu	1×	$0.05 / 1M prompt
PaddleOCR v5 Mobile paddleocr-v5-mobile · onnx-ocr	cpu_time	Ignite compute rate$0.0432 / vCPU-hr + $0.0060 / GB-RAM-hr

Vision detection · 5

Model	ID	Mode / Multiplier
MM Grounding DINO Large (GPU) mm-gdino-large · runpod-gpu	1×	$0.05 / 1M prompt
YOLOv8m Object Detection yolov8m · onnx-detection	cpu_time	Ignite compute rate$0.0432 / vCPU-hr + $0.0060 / GB-RAM-hr
SCRFD 10G Face Detection scrfd-10g · onnx-detection	cpu_time	Ignite compute rate$0.0432 / vCPU-hr + $0.0060 / GB-RAM-hr
SCRFD 34G Face Detection scrfd-34g · onnx-detection	cpu_time	Ignite compute rate$0.0432 / vCPU-hr + $0.0060 / GB-RAM-hr
PP-DocLayout-S Document Layout pp-doclayout-s · onnx-detection	cpu_time	Ignite compute rate$0.0432 / vCPU-hr + $0.0060 / GB-RAM-hr

Vision classification · 3

Model	ID	Mode / Multiplier
CLIP ViT-B/32 clip-vit-b32 · onnx-clip	cpu_time	Ignite compute rate$0.0432 / vCPU-hr + $0.0060 / GB-RAM-hr
MobileNetV3-Large (INT8) mobilenet-v3-large · onnx-classification	cpu_time	Ignite compute rate$0.0432 / vCPU-hr + $0.0060 / GB-RAM-hr
YAMNet Audio Classifier yamnet · onnx-classification	cpu_time	Ignite compute rate$0.0432 / vCPU-hr + $0.0060 / GB-RAM-hr

Text classification · 2

Model	ID	Mode / Multiplier	Effective rate
DistilBERT SST-2 distilbert-sst2 · onnx-text-encoder	cpu_time	Ignite compute rate$0.0432 / vCPU-hr + $0.0060 / GB-RAM-hr
Toxic-BERT (INT8) toxic-bert · onnx-text-encoder	cpu_time	Ignite compute rate$0.0432 / vCPU-hr + $0.0060 / GB-RAM-hr