Model Benchmarks & Unit Economics
Compare throughput, context scalability, and cost efficiency across frontier LLM architectures.
Frontier Model Matrix
Updated August 2026| Model | Provider | Throughput | Context Window | Input / 1M | Output / 1M | Reasoning |
|---|---|---|---|---|---|---|
| Claude 3.7 Sonnet | Anthropic | 68 tok/s | 200K | $3.00 | $15.00 | Supported |
| GPT-4o Omnimodal | OpenAI | 85 tok/s | 128K | $2.50 | $10.00 | — |
| DeepSeek R1 | DeepSeek | 42 tok/s | 128K | $0.55 | $2.19 | Supported |
| Gemini 2.5 Pro | 75 tok/s | 1M+ | $1.25 | $5.00 | Supported | |
| Llama 3.3 70B Turbo | Meta | 120 tok/s | 128K | $0.40 | $0.40 | — |
Interactive Cost & Latency Simulator
Estimate blended costs for batch inference workloads.
Input Prompt Tokens5,000
Output Generated Tokens1,500
Claude 3.7 Sonnet
Cost:$0.03750
Est Time:22.1s
GPT-4o Omnimodal
Cost:$0.02750
Est Time:17.6s
DeepSeek R1
Cost:$0.00604
Est Time:35.7s
Gemini 2.5 Pro
Cost:$0.01375
Est Time:20.0s
Llama 3.3 70B Turbo
Cost:$0.00260
Est Time:12.5s
Dynamic SVG Status Badges API
Embed live dynamic uptime and throughput badges directly into your documentation, Coolify dashboard, or GitHub readmes.
Claude 3.7 Sonnet
GPT-4o Omnimodal
DeepSeek R1
Gemini 2.5 Pro
Llama 3.3 70B Turbo