Compare AI models

Specifications
Parameters
—27B/35B/122B
Context window
1.05M256k
API pricing
Input price
$2.50—
Output price
$15.00—
Cached input price
$0.25—
Cheapest input
$1.25OpenAI—
Cheapest output
$7.50OpenAI—
$1,250OpenAI list price—
Benchmarks
Terminal-Bench 2.0
75.1%49.4%
BrowseComp
82.7%63.8%
Humanity's Last Exam · no tools
39.8%25.3%
Humanity's Last Exam · with tools
52.1%47.5%
GPQA Diamond
92.8%86.6%
OSWorld-Verified
75%58%
MMMU-Pro
81.2%76.9%
Overview
CompanyOpenAIQwen
Release dateMar 5 2026Feb 24 2026
AccessClosedOpen Weight
Model detailsView modelView model

Frequently asked questions

GPT-5.4 leads Qwen3.5 Medium on 7 of the 7 benchmarks they both report. Only GPT-5.4 has a verified first-party price for an available API: $2.50 per million input tokens and $15.00 per million output tokens. No currently available pay-as-you-go API rate is tracked for Qwen3.5 Medium. Qwen3.5 Medium shipped 9 days before GPT-5.4, so benchmark comparisons should account for the intervening progress.

Context windows are 1.05M (GPT-5.4) vs 256k (Qwen3.5 Medium). GPT-5.4 is closed, while Qwen3.5 Medium is open weight.

On Terminal-Bench 2.0, GPT-5.4 leads at 75.1% vs Qwen3.5 Medium at 49.4%. On BrowseComp, GPT-5.4 leads at 82.7% vs Qwen3.5 Medium at 63.8%. On Humanity's Last Exam · no tools, GPT-5.4 leads at 39.8% vs Qwen3.5 Medium at 25.3%. On Humanity's Last Exam · with tools, GPT-5.4 leads at 52.1% vs Qwen3.5 Medium at 47.5%. On GPQA Diamond, GPT-5.4 leads at 92.8% vs Qwen3.5 Medium at 86.6%. On OSWorld-Verified, GPT-5.4 leads at 75% vs Qwen3.5 Medium at 58%. On MMMU-Pro, GPT-5.4 leads at 81.2% vs Qwen3.5 Medium at 76.9%.