OpenAI's flagship multimodal model balancing speed, cost and intelligence.

Key specifications

Intelligence
77.5
Context window
128K
Max output
16.4K
Output speed
110 t/s
Latency (TTFT)
0.42s
Input $/1M
$2.50
Output $/1M
$10.00
License
Proprietary
Architecture
Transformer

Capabilities

Function callingVisionAudioStreamingJSON modeFine-tuning

Strengths

  • + Strong multimodal
  • + Fast
  • + Wide ecosystem

Limitations

  • Proprietary / closed weights
  • Pricier than open models

Benchmark scores

BenchmarkScore
GPQA Diamond53.6%
MMLU-Pro74.7%
SWE-Bench Verified33.2%
HumanEval90.2
AIME 202413.4%
GPT-4o — Specs, Pricing & Benchmarks | SyncDev