GPT-5.4 vs Claude Sonnet 4.6
Side-by-side comparison of API pricing, specs, benchmarks, and capabilities
| Specification |
by OpenAI
|
by Anthropic
|
|---|---|---|
| Overview | ||
| Status | Current flagship | Current mid tier |
| Released | Aug 15, 2025 | Jan 15, 2026 |
| Pricing per million tokens | ||
| Input | $2.50/Mtok | $3.00/Mtok |
| Output | $15.00/Mtok | $15.00/Mtok |
| Blended avg | $5.63/Mtok | $6.00/Mtok |
| Cached input | $0.250/Mtok | $0.300/Mtok |
| Specifications | ||
| Context window | 1M tokens | 1M tokens |
| Parameters | Proprietary | Proprietary |
| Speed (TPS) | — | — |
| Modalities | ||
| Input |
textimage
|
textimage
|
| Benchmarks sources: OpenAI model card, Artificial Analysis / Vellum LLM Leaderboard | ||
| Avg benchmark score | 78.7 | 70.4 |
| Perf / dollar | 14 | 11.7 |
| GPQA Diamond | 88 | 89.9 |
| Humanity's Last Exam | 40.2 | 19.1 |
| ARC-AGI 2 | 50 | 58.3 |
| AIME 2025 | 95.5 | 83 |
| MMMLU | 88 | 89.3 |
| HumanEval | 94.5 | 79.6 |
| SWE-Bench Verified | 75 | — |
| Terminal-Bench 2.1 | — | 59.1 |
| LiveCodeBench | — | 72.4 |
| MCP Atlas | — | 69.5 |
| BrowseComp | — | 76.2 |
| OSWorld-Verified | — | 78.5 |
| BFCL | 85 | — |
| MATH 500 | 92 | — |
| AutoBench | — | 5.3 |
| Independent composite scores each on its own scale — not part of the average above | ||
| AA Intelligence Index | 53.1 | 48.4 |
| AA Agentic Index | 44.2 | 42.1 |
| AA-Omniscience Index (−100–100) | 5.8 | 12.2 |
| LMSYS Chatbot Arena | 1465.5 ELO | 1472 ELO |
| Providers | ||
| Available from |
OpenAI — $2.50/$15.00/Mtok
|
Anthropic — $3.00/$15.00/Mtok
|
Cost at scale
1M tokens · 50/50 input/output| Volume | GPT-5.4 | Claude Sonnet 4.6 | Savings |
|---|---|---|---|
| 1M tokens | $5.63 | $6 | $0.38 (6.3%) |
| 10M tokens | $56.25 | $60 | $3.75 (6.3%) |
| 100M tokens | $562.5 | $600 | $37.5 (6.3%) |
| 1000M tokens | $5625 | $6000 | $375 (6.3%) |
When to pick which
distilled from the pricing and spec data above- cost dominates: $5.63/Mtok blended vs $6.00 — 6.3% less on the same 50/50 token mix
- raw capability matters more than price: 78.7 vs 70.4 average benchmark score
- this pairing gives Claude Sonnet 4.6 no edge on price, context, cache discount, or measured performance — pick it only for qualitative fit (ecosystem, compliance, model behavior on your prompts)
The cheaper option here — GPT-5.4 costs $5.63/Mtok blended on OpenAI.
Summary
GPT-5.4 by OpenAI costs $2.50/Mtok input and $15.00/Mtok output, with a 1M-token context window. It supports text, image input.
Claude Sonnet 4.6 by Anthropic costs $3.00/Mtok input and $15.00/Mtok output, with a 1M-token context window. It supports text, image input.
On a blended cost basis, GPT-5.4 is 6.3% cheaper than Claude Sonnet 4.6.
On benchmarks, GPT-5.4 scores higher (78.7 vs 70.4) on average. In terms of value, GPT-5.4 has better performance per dollar (14 vs 11.7).
Note: Pricing is per million tokens. Actual costs vary with usage patterns, prompt caching, and batch discounts. Always verify against official provider pricing pages.