Compare / head-to-head
GPT-5.3-CodexvsGPT-5.5
GPT-5.5 leads 5 of 5 shared benchmarks. GPT-5.3-Codex is 2.2x cheaper per token. GPT-5.5 has the larger context window (1M tokens).
Benchmarks from cited public sources; pricing from official pages; status from official provider feeds.
Shared benchmarks
0 – 5
GPT-5.5 leads
Cheaper per token
GPT-5.3-Codex
2.2x cheaper, input + output
Larger context
GPT-5.5
1M tokens
Providers
OpenAI
same provider
GPT-5.3-Codex
OpenAI · released 2026-02-05
textvision
Quality
Benchmark matrix
5 shared · 4 only GPT-5.3-Codex · 39 only GPT-5.5| Benchmark | GPT-5.3-Codex | GPT-5.5 | Δ | Edge |
|---|---|---|---|---|
| Reported by both · 5 | ||||
| GDPval (wins or ties) | 70.9% ↗ | 84.9% ↗ | -14 pt | GPT-5.5 |
| OSWorld-Verified | 64.7% ↗ | 78.7% ↗ | -14 pt | GPT-5.5 |
| Terminal-Bench 2.0 | 77.3% ↗ | 82.7% ↗ | -5.4 pt | GPT-5.5 |
| Vending-Bench 2 (Andon Labs) | 5940.12 ↗ | 7523.84 ↗ | -1583.7 | GPT-5.5 |
| WeirdML (Håvard Tveit Ihle) | 79.3% ↗ | 84.9% ↗ | -5.6 pt | GPT-5.5 |
| Only GPT-5.3-Codex reports · 4 | ||||
| Cybersecurity Capture The Flag Challenges | 77.6% ↗ | not reported | — | — |
| SWE-Bench Pro (Public) | 56.8% ↗ | not reported | — | — |
| SWE-bench Verified (Epoch AI run) | 74.8% ↗epoch run | not reported | — | — |
| SWE-Lancer IC Diamond | 81.4% ↗ | not reported | — | — |
| Only GPT-5.5 reports · 39 | ||||
| GPQA Diamond | not reported | 90.7% ↗epoch run | — | — |
| AIME 2026 | not reported | 100% ↗matharena ⚠ | — | — |
| LMArena Elo | not reported | 1465.6 ↗ | — | — |
| APEX-Agents (Mercor) | not reported | 55.1% ↗ | — | — |
| ARC-AGI-1 (Verified) | not reported | 95% ↗ | — | — |
| ARC-AGI-2 (Verified) | not reported | 85% ↗ | — | — |
| BrowseComp | not reported | 84.4% ↗ | — | — |
| Chess Puzzles (Epoch AI run) | not reported | 26% ↗epoch run | — | — |
| DeepSWE v1.1 (Datacurve) | not reported | 67% ↗ | — | — |
| EBR-bench (Epoch AI run) | not reported | 34.3% ↗epoch run | — | — |
| FrontierMath Tier 1-3 | not reported | 51.7% ↗ | — | — |
| FrontierMath Tier 4 | not reported | 35.4% ↗ | — | — |
| FrontierMath Tier 4 v2 (Epoch AI run) | not reported | 72.5% ↗epoch run | — | — |
| FrontierMath Tiers 1-3 v2 (Epoch AI run) | not reported | 85.3% ↗epoch run | — | — |
| Furniture Assembly (Epoch AI run) | not reported | 44.2% ↗epoch run | — | — |
| GSO Opt@1 (GSO) | not reported | 37.3% ↗ | — | — |
| LiveBench Agentic Coding (LiveBench) | not reported | 54% ↗ | — | — |
| LiveBench Coding (LiveBench) | not reported | 82.1% ↗ | — | — |
| LiveBench Data Analysis (LiveBench) | not reported | 81.6% ↗ | — | — |
| LiveBench Instruction Following (LiveBench) | not reported | 70.7% ↗ | — | — |
| LiveBench Language (LiveBench) | not reported | 87.4% ↗ | — | — |
| LiveBench Mathematics (LiveBench) | not reported | 95.9% ↗ | — | — |
| LiveBench Reasoning (LiveBench) | not reported | 89.7% ↗ | — | — |
| LMArena Agent (LMArena) | not reported | 0.0415 ↗ | — | — |
| LMArena Vision (LMArena) | not reported | 1286.5 ↗ | — | — |
| MCP Atlas | not reported | 75.3% ↗ | — | — |
| MirrorCode (Epoch AI run) | not reported | 10% ↗epoch run | — | — |
| MMMU-Pro (no tools) | not reported | 81.2% ↗ | — | — |
| MMMU-Pro (with tools) | not reported | 83.2% ↗ | — | — |
| Mystery Game Puzzles (Epoch AI run) | not reported | 56% ↗epoch run | — | — |
| OTIS Mock AIME 2024-2025 (Epoch AI run) | not reported | 84.4% ↗epoch run | — | — |
| SAGE (Vals AI) | not reported | 51.5% ↗ | — | — |
| SimpleBench (SimpleBench) | not reported | 69% ↗ | — | — |
| SimpleQA Verified | not reported | 63% ↗epoch run | — | — |
| SWE-Bench Pro | not reported | 58.6% ↗ | — | — |
| tau2-bench Banking Knowledge (Sierra) | not reported | 44.6% ↗ | — | — |
| Tau2-bench Telecom | not reported | 98% ↗ | — | — |
| Toolathlon | not reported | 55.6% ↗ | — | — |
| Toolathlon-Verified (HKUST) | not reported | 73.5% ↗ | — | — |
Scores tagged "epoch" or "matharena" are independent runs, used only where the lab has not published its own; ⚠ marks rows MathArena flags as released after the competition. Higher is better on every row. Δ is GPT-5.3-Codex minus GPT-5.5 in the benchmark's own unit. "Not reported" means the lab has not published that figure; it is not a zero. ↗ opens the source.
Specs & pricing
Side by side
Official pricing pages and model cards| Spec | GPT-5.3-Codex | GPT-5.5 | Edge |
|---|---|---|---|
| Context window | 400K tokens | 1M tokens | GPT-5.5 |
| Max output | 128K tokens | 128K tokens | Tie |
| Input price / 1M | $1.75 ↗ | $5 ↗ | GPT-5.3-Codex |
| Output price / 1M | $14 ↗ | $30 ↗ | GPT-5.3-Codex |
| Cached input / 1M | $0.175 ↗ | $0.5 ↗ | GPT-5.3-Codex |
| Input + output / 1M Lower is cheaper. List prices; batch, tool and regional fees excluded. | $15.75 | $35 | GPT-5.3-Codex |
| Modalities | text · vision | text · vision | Tie |
| Released | 2026-02-05 | 2026-04-23 | — |
| Cited benchmark scores | 9 | 50 | — |
Reliability
OpenAI status
Live from /statusMore matchups
GPT-5.3-Codex vs …
Models sharing the most benchmarksMore matchups
GPT-5.5 vs …
Models sharing the most benchmarksBuilt by Respan
Which one wins on your data?
Public benchmarks are a starting point. Run GPT-5.3-Codex and GPT-5.5 on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.