Models

Browse Models

Fastlane routes your requests to the best model for each task. Below are all models available across BYOK providers, OpenRouter, and Marketplace.

Model Comparison

Model	Provider	Context	MMLU	HumanEval	Price (Input)	Best For
GPT-4o	OpenAI	128k	88.7%	90.2%	$2.50/1M	Complex reasoning, coding
Claude Sonnet 4	Anthropic	200k	88.7%	92.0%	$3.00/1M	Analysis, writing, coding
Llama 3.3 70B	Groq	128k	86.0%	89.0%	$0.59/1M	Open-source quality, fast
GPT-4o Mini	OpenAI	128k	82.0%	87.0%	$0.15/1M	General tasks, quick answers
Mixtral 8x7B	Groq	32k	77.3%	74.4%	$0.27/1M	Balanced open-source
Claude 3.5 Haiku	Anthropic	200k	75.2%	75.9%	$1.00/1M	Fast responses, summarization
Gemma 2 9B	Groq	8k	71.3%	59.1%	$0.20/1M	Ultra-fast, simple tasks
Llama 3.1 8B	Groq	128k	68.4%	62.0%	$0.20/1M	Quick tasks, high volume

Scores from official technical reports and LMSYS Chatbot Arena

Performance Tiers

High Quality (MMLU 85%+)

• GPT-4o, Claude Sonnet 4, Llama 3.3 70B
• Best for: Complex coding, analysis, multi-step reasoning
• Recommended modes: Accurate, Balanced

Medium Quality (MMLU 65-85%)

• GPT-4o Mini, Mixtral 8x7B, Claude 3.5 Haiku, Gemma 2 9B, Llama 3.1 8B
• Best for: General tasks, quick answers, classification
• Recommended modes: Cheap, Balanced, Eco

How Routing Works

When you send a request with fastlane-auto, Fastlane:

1. Analyzes your prompt
2. Scores every available model using the ONNX ranking model
3. Picks the best model based on your mode (Eco, Cheap, Balanced, Accurate)
4. Routes the request and returns the response

You can also send requests directly to any model by name.

BYOK vs Marketplace

Feature	BYOK	Marketplace
Requires API key	Yes	No
Pricing	Provider rates	3% markup
Model selection	Your keys only	All models
Affiliation	N/A	1% commission

Pricing

All prices shown are per 1 million input tokens. Output tokens are typically 2-4x more expensive.

Prices include the 3% Marketplace markup where applicable. BYOK prices are direct from the provider.