OpenAI's high-capability GPT-4 variant with 128K context — now superseded by GPT-4o
OpenAI no longer offers this model (shut down 2026-10-23). Recommended replacement: gpt-6-1-sol.
Checked 2026-11-12 against OpenAI's documentation
| Provider | OpenAI |
| Released | 2023-11-06 |
| Status | Retired on 2026-10-23 |
| Context window | 128K tokens |
| Pricing | Paid |
| Input price | $10.00/M |
| Output price | $30.00/M |
| Capabilities | textvisioncodefunction-calling |
| API docs | Docs ↗ |
| Replaced by | gpt-6-1-sol |
| GPT-4 Turbo | Mistral Large 2 | Gemini 2.5 Pro | Claude Opus 4.6 | |
|---|---|---|---|---|
| Context | 128K | 131K | 1.05M | 200K |
| MMLU | 86.4 | 84.0 | 91.0 | 92.0 |
| HumanEval | 87.1 | 92.0 | — | 94.5 |
| MATH | — | 69.7 | 97.0 | 97.1 |
| GPQA | — | — | 84.0 | 79.9 |
| Pricing | Paid | Paid | Paid | Paid |
| Input $/M | $10.00 | $2.00 | $1.25 | $15.00 |
GPT-4 Turbo extended the original GPT-4 with a 128K context window, vision support, and lower pricing. It was the recommended high-capability option through most of 2024, but has since been superseded by GPT-4o which matches its intelligence benchmarks at lower cost and higher speed.
For new projects, use GPT-4o instead. GPT-4o matches GPT-4 Turbo on most tasks, runs 2× faster, and costs significantly less per token.