OpenAI · Released Apr 24, 2026
GPT-5.5
GPT-5.5 is OpenAI’s frontier model designed for complex professional workloads, building on GPT-5.4 with stronger reasoning, higher reliability, and improved token efficiency on hard tasks. It features a 1M+ token...
- #1Terminal-Bench
- Top 9%Chess Puzzles
- #1ExploitBench
- Top 14%CritPt
- Released
- Apr 24 2026
- Context window
- 1.05M 128K out
- Input / 1M tokens
- $5.00 from $2.50
- Output / 1M tokens
- $30.00 from $15.00
Where to run it
3 providers compared
Cheapest: OpenAI at $2.50 / $15.00 per 1M input / output tokens — 80% less than the priciest option.
| Provider | Input | Output | Cache read | Context | Max output | Uptime | vs. best |
|---|---|---|---|---|---|---|---|
| OpenAI | $2.50 | $15.00 | $0.25 | 1.05M | 128K | 100%24h | Cheapest |
| Azure | $5.00 | $30.00 | $0.50 | 1.05M | 128K | 100%24h | +100% |
| OpenAI | $5.00 | $30.00 | $0.50 | 1.05M | 128K | 99.9%24h | +100% |
| Azureus | $5.50 | $33.00 | $0.55 | 1.05M | 128K | 100%24h | +120% |
| Azureeu | $5.50 | $33.00 | $0.55 | 1.05M | 128K | 100%24h | +120% |
| Amazon Bedrockus-east-1 | $5.50 | $33.00 | $0.55 | 1.05M | 128K | — | +120% |
| OpenAIfast | $12.50 | $75.00 | $1.25 | 1.05M | 128K | — | +400% |
Value
Price vs. performance
GPT-5.5 Other OpenAI Other labs Value frontier · GPQA Diamond vs. input price per 1M tokens (log scale) · Drag to pan, pinch or Ctrl + scroll to zoom, double-click to zoom in.
Benchmarks
30 benchmark scores
| Benchmark | Score | Rank | Leader |
|---|---|---|---|
| DeepSWESoftware engineering · Original long-horizon tasks across active open-source repos | 67.0% | #11of 26 | GPT-6 Astra74.1% |
| FrontierCodeSoftware engineering · Mergeable fixes for hard open-source issues | 43.0% | #19of 41 | Claude Opus 5.554.6% |
| Terminal-BenchSoftware engineering · Tasks completed with the programs in a real terminal (best agent) | 84.7% | #1of 40 | This model |
| MirrorCodeSoftware engineering · Reimplement whole programs end-to-end without the source | 10.0% | #8of 9 | Claude Opus 5.577.4% |
| GBAEvalSoftware engineering · Build a Game Boy Advance emulator from scratch | 53.2% | #6of 23 | Claude Opus 579.6% |
| GSOSoftware engineering · Optimize real code to make it significantly faster | 40.2% | #8of 20 | Claude Fable 5.188.2% |
| SciCodeSoftware engineering · Research-coding problems from the natural sciences | 56.1% | #23of 93 | Claude Opus 5.566.9% |
| APEX-AgentsAgents & computer use · Long professional tasks in banking, consulting and law | 55.1% | #18of 51 | Gemini 4 Argon82.4% |
| Remote Labor IndexAgents & computer use · Real paid freelance projects done end-to-end | 6.3% | #6of 13 | Claude Opus 5.521.3% |
| OSWorld 2.0Agents & computer use · Long real-world desktop and web tasks | 13.0% | #5of 9 | Claude Opus 531.4% |
| Vending-Bench 2Agents & computer use · Run a simulated vending business for a year (final balance) | $7,524 | #14of 61 | GPT-6 Astra$15,515 |
| PostTrainBenchAgents & computer use · Post-train small models under a fixed compute budget | 27.2% | #8of 11 | Claude Fable 541.8% |
| DeepResearch BenchWriting & content · Find and synthesize information from the web | 54.0% | #4of 20 | Claude Opus 4.655.3% |
| GDP.pdfWriting & content · Answer questions about real professional PDFs | 26.0% | #11of 38 | GPT-6.1 Sol35.0% |
| SimpleQA VerifiedWriting & content · Factual accuracy on 1,000 short questions | 63.0% | #14of 66 | Gemini 4 Argon76.0% |
| WebDev ArenaDesign & visual · People vote on which model builds the better website (Elo) | 1510 | #43of 88 | Claude Opus 5.51820 |
| Blueprint-Bench 2Design & visual · Turn apartment photos into 2D floor plans | 36.2% | #9of 32 | Gemini 4 Argon54.4% |
| Furniture AssemblyDesign & visual · Spot mistakes in photos of IKEA builds | 44.2% | #12of 34 | Claude Opus 5.583.3% |
| ExploitBenchCybersecurity · Exploit real, hardened security vulnerabilities | 47.4% | #1of 8 | This model |
| GPQA DiamondReasoning & knowledge · PhD-level biology, chemistry and physics questions | 90.7% | #33of 92 | Gemini 4 Argon96.4% |
| ARC-AGI-2Reasoning & knowledge · Few-shot abstract reasoning on grid puzzles | 85.0% | #11of 63 | GPT-6 Astra95.0% |
| ARC-AGI-1Reasoning & knowledge · Pattern induction from input–output examples | 95.0% | #15of 63 | Gemini 4 Argon98.5% |
| SimpleBenchReasoning & knowledge · Common-sense trick questions about space, time and people | 69.0% | #12of 43 | Claude Fable 581.9% |
| CritPtReasoning & knowledge · Open-ended, research-level physics problems | 27.1% | #13of 96 | GPT-5.6 Sol32.3% |
| Chess PuzzlesReasoning & knowledge · Find the single best move in 100 new puzzles | 63.0% | #6of 70 | GPT-6 Astra76.0% |
| LMCAReasoning & knowledge · Judge the quality of conceptual arguments like experts | 54.3% | #15of 96 | Claude Sonnet 5.568.3% |
| FrontierMathMaths · Expert-written problems, undergrad to early research (tiers 1–3) | 85.3% | #13of 72 | GPT-6.1 Sol93.7% |
| FrontierMath Tier 4Maths · Exceptionally hard research-level problems | 72.5% | #12of 62 | GPT-6.1 Sol100.0% |
| ProofBenchMaths · Graduate maths proofs verified in Lean 4 | 50.0% | #31of 70 | Claude Sonnet 5.5100.0% |
| OTIS Mock AIMEMaths · Competition-style problems harder than MATH Level 5 | 84.4% | #66of 90 | Gemini 4 Argon100.0% |
Specs
Details
- Model ID
openai/gpt-5.5- Developer
- OpenAI
- Release date
- Apr 24, 2026 ·
- Context window
- 1,050,000 tokens
- Max output
- 128,000 tokens
- API price
- $5.00 / $30.00 per 1M input / output tokens
- Providers
- 3
- Input types
- file, image, text
- Reasoning
- Yes
- Tool calling
- Yes
- Open weights
- No
- Free tier
- No
Release date, prices and providers from OpenRouter's catalog; buying directly from a provider can cost a little differently. Uptime is the share of OpenRouter requests each provider served over the last day. Benchmark scores are the best result per benchmark from Epoch AI, licensed CC BY 4.0. Rank counts every model tracked here with a score.