GPT-5.4 2026-03-05
GPT-5.4 is OpenAI’s latest frontier model, unifying the Codex and GPT lines into a single system. It features a 1M+ token context window (922K input, 128K output) with support for text and image inputs, enabling high-context reasoning, coding, and multimodal analysis within the same workflow. The model delivers improved performance in coding, document understanding, tool use, and instruction following. It is designed as a strong default for both general-purpose tasks and software engineering, capable of generating production-quality code, synthesizing information across multiple sources, and executing complex multi-step workflows with fewer iterations and greater token efficiency.
- Input / output price
- $2.50 / $15.00 per 1M
- Context
- 1.1M tokens
- Released
- Mar 5, 2026
Paste into Claude Code, Cursor, or Codex
Help me set up the Respan gateway: fetch and follow https:/
Providers
One provider serves this model, priced per 1M tokens after gateway discounts.
| Provider | Input | Output |
|---|---|---|
OpenAI openai/gpt-5.4-2026-03-05 | $2.50/M | $15.00/M |
Is a provider down? ChatGPT status
Pricing
USD per 1M tokens on OpenAI, after gateway discounts.
- Input
- $2.50 per 1M
- Output
- $15.00 per 1M
- Cache read
- $0.25 per 1M
Cost by volume
Total cost from 0 to 1T tokens.
Throughput and latency
30-day averages from requests through the Respan gateway. Under each figure, every model on the gateway by its average, this one in blue.
Output speed
Higher is better28tokens/sacross providers
Response latency
Lower is better1.3sacross providers
Benchmarks
The Respan Index plus 43 published scores from 12 sources, each linked to where it was published.
50.0/ 100
Respan Index#43 of 176
- 43GPT-5.450.0
- 42GPT-5.4 Pro51.0
- 43Claude Opus 4.750.0
Published scores
| Benchmark | Score | Source | Published |
|---|---|---|---|
| WeirdMLHåvard Tveit Ihle | 77.7% | Håvard Tveit Ihle | Oct 5, 2026 |
| LiveBench CodingLiveBench | 77.5% | LiveBench | Oct 5, 2026 |
| SWE-bench VerifiedEpoch AI | 76.9% | Epoch AI | Oct 5, 2026 |
| Terminal-Bench 2.0OpenAI | 75.1% | OpenAI | Mar 5, 2026 |
| SWE-Bench ProOpenAI | 57.7% | OpenAI | Mar 5, 2026 |
| DeepSWE v1.1Datacurve | 51.8% | Datacurve | Oct 5, 2026 |
| GSO Opt@1GSO | 30.4% | GSO | Mar 10, 2026 |
| MirrorCodeEpoch AI | 23.3% | Epoch AI | Nov 11, 2026 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| Tau2-bench TelecomOpenAI | 98.9% | OpenAI | Mar 5, 2026 |
| GDPval (wins or ties)OpenAI | 83% | OpenAI | Mar 5, 2026 |
| BrowseCompOpenAI | 82.7% | OpenAI | Mar 5, 2026 |
| OSWorld-VerifiedOpenAI | 75% | OpenAI | Mar 5, 2026 |
| MCP AtlasOpenAI | 67.2% | OpenAI | Mar 5, 2026 |
| ToolathlonOpenAI | 54.6% | OpenAI | Mar 5, 2026 |
| LiveBench Agentic CodingLiveBench | 53.8% | LiveBench | Oct 5, 2026 |
| APEX-AgentsMercor | 52.4% | Mercor | Oct 5, 2026 |
| tau2-bench Banking KnowledgeSierra | 39.4% | Sierra | Mar 25, 2026 |
| Vending-Bench 2Andon Labs | 6144.18 · #27 of 68 | Andon Labs | Oct 5, 2026 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| GPQA DiamondEpoch AI | 93.3% | Epoch AI | Oct 5, 2026 |
| GPQA DiamondOpenAI | 92.8% | OpenAI | Mar 5, 2026 |
| LiveBench ReasoningLiveBench | 88.1% | LiveBench | Oct 5, 2026 |
| LiveBench Data AnalysisLiveBench | 79.3% | LiveBench | Oct 5, 2026 |
| Humanity's Last Exam (with tools)OpenAI | 52.1% | OpenAI | Mar 5, 2026 |
| Chess PuzzlesEpoch AI | 48% | Epoch AI | Oct 28, 2026 |
| Humanity's Last Exam (no tools)OpenAI | 39.8% | OpenAI | Mar 5, 2026 |
| Mystery Game PuzzlesEpoch AI | 37% | Epoch AI | Oct 5, 2026 |
| EBR-benchEpoch AI | 25.4% | Epoch AI | Oct 5, 2026 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| AIME 2026MathArena | 99.2% | MathArena | Sep 17, 2026 |
| OTIS Mock AIME 2024-2025Epoch AI | 97.8% | Epoch AI | Oct 5, 2026 |
| LiveBench MathematicsLiveBench | 94.1% | LiveBench | Oct 5, 2026 |
| FrontierMath Tiers 1-3 v2Epoch AI | 78.6% | Epoch AI | Oct 5, 2026 |
| FrontierMath Tier 4 v2Epoch AI | 49% | Epoch AI | Oct 5, 2026 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| LiveBench LanguageLiveBench | 82.6% | LiveBench | Oct 5, 2026 |
| LiveBench Instruction FollowingLiveBench | 70.2% | LiveBench | Oct 5, 2026 |
| SimpleQA VerifiedEpoch AI | 45.1% | Epoch AI | Oct 5, 2026 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| MMMU-Pro (with tools)OpenAI | 82.1% | OpenAI | Mar 5, 2026 |
| MMMU-Pro (no tools)OpenAI | 81.2% | OpenAI | Mar 5, 2026 |
| Furniture AssemblyEpoch AI | 37.5% | Epoch AI | Oct 5, 2026 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| LMArena EloLMArena | 1475 Elo · #35 of 191 | LMArena | Nov 19, 2026 |
| LMArena WebDevLMArena | 1462 Elo · #72 of 124 | LMArena | Nov 20, 2026 |
| LMArena VisionLMArena | 1282 Elo · #20 of 91 | LMArena | Nov 19, 2026 |
| LMArena AgentLMArena | 0.0267 · #23 of 68 | LMArena | Oct 12, 2026 |
How GPT compares: Claude vs GPT, Gemini vs GPT, DeepSeek vs GPT
Compare GPT-5.4 2026-03-05: GPT-5.4 vs GPT-5.2, GPT-5.4 vs GPT-5.5, GPT-5.4 vs GPT-6.1 Sol, GPT-5.4 vs GPT-6.2 Astra, GPT-5.4 vs GPT-6 Astra, GPT-5.4 vs GPT-6.1 Terra
Apps
Use GPT-5.4 2026-03-05 in the tools you already run. Each guide points the app at the Respan gateway.
Setup guides for more apps are in the integration docs.
GPT-5.4 2026-03-05 FAQ
Short answers, built from the data on this page.
GPT-5.4 2026-03-05 is a model from OpenAI, released on Mar 5, 2026. GPT-5.4 is OpenAI’s latest frontier model, unifying the Codex and GPT lines into a single system. It features a 1M+ token context window (922K input, 128K output) with support for text and image inputs, enabling high-context reasoning, coding, and multimodal analysis within the same workflow. The model delivers improved performance in coding, document understanding, tool use, and instruction following. It is designed as a strong default for both general-purpose tasks and software engineering, capable of generating production-quality code, synthesizing information across multiple sources, and executing complex multi-step workflows with fewer iterations and greater token efficiency.
On the Respan gateway, GPT-5.4 2026-03-05 costs $2.50 per 1M input tokens and $15.00 per 1M output tokens on OpenAI, after gateway discounts.
GPT-5.4 2026-03-05 is served by OpenAI.
GPT-5.4 2026-03-05 accepts up to 1.1M tokens of context.
Point any OpenAI-compatible SDK at https://gateway.respan.ai/v1 with your Respan API key and set the model to "openai/gpt-5.4-2026-03-05". To set it up with a coding agent, copy the agent setup prompt at the top of this page.
GPT-5.4 2026-03-05 scores 50.0 out of 100 on the Respan Index v1.1, ranking #43 of 176 models. By category: Coding 25.9, Agents 62.0, Reasoning 61.4, Knowledge & language 46.9, and Human preference 52.9. The Respan Index combines independent and lab-reported benchmark results into one score, using benchmark data collected by Respan.
GPT-5.4 2026-03-05 has 43 published benchmark scores, including WeirdML 77.7%, Tau2-bench Telecom 98.9%, and GPQA Diamond 93.3%. Each score links to where it was published; the data is collected by Respan.
Certified where it counts.
Built to meet the security and privacy standards that enterprise and healthcare teams require.
ISO 27001
The internationally recognized standard for information security management.
SOC 2
Secure, compliant management of your data across all of our systems.
GDPR
Operated under GDPR, the world's strictest standard for data privacy.
HIPAA
HIPAA compliant, with a BAA available for healthcare teams.