CallMissed
Back to all models
LLM Chatrecommended

GPT-5.4

by OpenAI · Released March 2026

OpenAI's flagship general-purpose model. Balances high capability with reasonable cost. Features 1M token context, native computer use, and improved accuracy for professional workflows including coding, writing, and analysis.

LLM Chat

GPT-5.4

Powered by OpenAI · Transformer (proprietary)

Context Window

1M

Parameters

Undisclosed

Max Output

32K

Category

LLM Chat

Overview

GPT-5.4 is OpenAI's unified frontier model, combining reasoning, coding, and computer use into a single system at a fraction of the cost of GPT-5.4 Pro. Priced at $2.50/M input and $15/M output from OpenAI (available at $3.50/$20 through CallMissed), it offers the same 1M token context window (272K standard, 1M in Codex experimental) and native computer use capabilities, making it the sweet spot of the GPT-5.4 family for most production workloads.

The model features tool search that loads tool definitions on demand instead of all at once, saving tens of thousands of tokens per request. Its native computer use interacts with desktops through screenshots, controls mouse and keyboard, and writes Playwright code for browser automation. On GDPval, it tops the leaderboard at 1667 Elo alongside the Pro variant, and human raters preferred GPT-5.4 presentations 68% of the time over GPT-5.2.

On coding benchmarks, GPT-5.4 scores 57.7% on SWE-bench Pro, 75.0% on Terminal-Bench 2.0, and a remarkable 75.0% on OSWorld-Verified — exceeding human performance of 72.4%. Web browsing is equally strong: 82.7% on BrowseComp, 67.3% on WebArena-Verified, and 92.8% on Online-Mind2Web. For reasoning, it achieves 47.6% on FrontierMath, 52.1% on Humanity's Last Exam (breaking the 50% threshold), and 93.7% on ARC-AGI-1 with 73.3% on ARC-AGI-2.

GPT-5.4 also delivers a 33% reduction in false claims and 18% fewer responses containing any errors. Its steerability is enhanced — it outlines a plan before continuing and allows mid-response adjustments. For teams that need frontier-level capabilities without the premium pricing of GPT-5.4 Pro, this is the recommended choice for complex multi-step tasks, code generation, content creation, and data analysis.

Pricing

MetricPrice
Input /1M tokens₹350.0000
Output /1M tokens₹2000.0000

1 credit = ₹1 = $0.01 USD. Prices shown from provider; CallMissed passes through with ~35% markup.

Key Highlights

  • Best price-to-performance ratio in the GPT-5.4 family
  • 1M token context window
  • Native computer use and tool search
  • Strong coding, math, and reasoning performance

Benchmarks

BenchmarkScore
SWE-bench Pro57.7%
Terminal-Bench 2.075.0%
OSWorld-Verified75.0%
BrowseComp82.7%
FrontierMath47.6%
Humanity's Last Exam52.1%
ARC-AGI-193.7%
ARC-AGI-273.3%
GDPval1667 Elo
Spreadsheet Modeling87.3%

Technical Details

  • Context window: 1,000,000 tokens (272K standard, 1M in Codex experimental)
  • Native computer use: interacts with desktop via screenshots, controls mouse/keyboard, writes Playwright code
  • Tool search: loads tool definitions on demand, saving tens of thousands of tokens per request
  • 33% fewer false claims and 18% fewer responses with any errors vs predecessors
  • Steerability: outlines plan before continuing, allows mid-response adjustments
  • Pricing: $2.50/M input, $15/M output from OpenAI directly
  • Supports structured outputs, function calling, and JSON mode
  • Post-trained with RLHF for instruction following and safety
  • Available via OpenAI API and CallMissed unified gateway

Strengths

  • Best price-to-performance in the GPT-5.4 family — frontier capability at reasonable cost
  • 1M context window handles entire codebases and long documents
  • Native computer use and tool search for agentic workflows
  • Strong across coding, reasoning, math, and general knowledge

Limitations

  • Slightly lower benchmark scores than GPT-5.4 Pro on professional tasks
  • Proprietary and closed-source — no self-hosting option
  • Higher cost than open-source alternatives like GPT-OSS-120B

Use Cases

General-purpose AI assistantCode generationContent creationData analysis

API Example

curl https://api.callmissed.com/v1/chat/completions \
  -H "Authorization: Bearer cm_YOUR_KEY" \
  -d '{"model": "openai/gpt-5.4", "messages": [{"role": "user", "content": "Write a Python FastAPI endpoint with proper error handling"}]}'

Endpoint: POST /v1/chat/completions · Model ID: openai/gpt-5.4

Try GPT-5.4 now

Get 1000 free API credits on signup. No credit card required.

Start free Read docs