Anthropic: Claude Sonnet 4.5 vs OpenAI: GPT-5
On input price, OpenAI: GPT-5 is ~2.4× cheaper. Verified 2026-10-21.
Anthropic: Claude Sonnet 4.5 $3.00/$15.00vsOpenAI: GPT-5 $1.25/$10.00
Verdict
On blended price (3:1 input:output), OpenAI: GPT-5 is ~1.7× cheaper ($1.25/$10.00 vs $3.00/$15.00 per 1M tokens). Anthropic: Claude Sonnet 4.5 generates 73% faster output.
Pricing
Per-million-token list prices, verified 2026-10-21. Green marks the better value.
| Metric | Anthropic: Claude Sonnet 4.5 | OpenAI: GPT-5 |
|---|---|---|
| Input / 1M | $3.00 | $1.25 |
| Output / 1M | $15.00 | $10.00 |
| Cache read / 1M | $0.300 | $0.125 |
| Blended (3:1) | $6.00/1M | $3.44/1M |
| Estimated monthly cost | Anthropic: Claude Sonnet 4.5 | OpenAI: GPT-5 |
|---|---|---|
| Chatbot10,000 reqs · 800/400 tok | $84.00 | $50.00 |
| Coding agent5,000 reqs · 8000/2000 tok | $270.00 | $150.00 |
| RAG pipeline30,000 reqs · 4000/500 tok | $585.00 | $300.00 |
Plug in your own numbers in the calculator, or paste a real bill into the Bill Shredder.
Specifications & capabilities
| Metric | Anthropic: Claude Sonnet 4.5 | OpenAI: GPT-5 |
|---|---|---|
| Context window | 1M | 400K |
| Max output | 64K | 128K |
| Vision input | ✓ | ✓ |
| Tool calling | ✓ | ✓ |
| Reasoning | ✓ | ✓ |
Performance
Measured by Artificial Analysis on live endpoints.
Output speed — higher is better
Anthropic: Claude Sonnet 4.5130 tok/s
OpenAI: GPT-575 tok/s
Latency (TTFT) — lower is better
Anthropic: Claude Sonnet 4.5700 ms
OpenAI: GPT-51,400 ms
| Metric | Anthropic: Claude Sonnet 4.5 | OpenAI: GPT-5 |
|---|---|---|
| Cost per task | $0.000008 | $0.000020 |
Provider
Infrastructure metrics for the organizations behind these models.
| Metric | Anthropic | OpenAI |
|---|---|---|
| Listed endpoints | 8 | 12 |
| Reliability | 99.8% | 99.9% |
| Avg latency | 520 ms | 450 ms |
| Pricing tiers | standard, batch | standard, batch, provisioned |
Full details: Anthropic: Claude Sonnet 4.5 pricing · OpenAI: GPT-5 pricing