OpenAI model product
GPT-5.5
5 is OpenAI's smartest model yet, designed for real work across agentic coding, computer use, knowledge work, and early scientific research.
Updated Aug 12, 2026. Default version: GPT-5.5
LLMBoard score84.0GPT-5.5
Coverage80%26 benchmark families
Context window1.1MTokens
Official input price$5OpenAI API
GPT-5.5 Specifications
Technical details for the model's default version.
- Version
- GPT-5.5
- Released
- Apr 23, 2026
- Knowledge cutoff
- Dec 1, 2025
- Parameters
- N/A
- Context window
- 1.1M
- Max output
- 128K
- Inputs
- image, pdf, text
- Outputs
- text
- Open weights
- No
- License
- Proprietary
GPT-5.5 Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
GPT-5.5 Benchmark Results
Benchmark scores for GPT-5.5.
| ARC-AGI | 95.0% | 01 | 7 | 100.0% | C | |
| ARC-AGI v2 | 85.0% | 01 | 16 | 100.0% | C | |
| BixBench | 80.5% | 01 | 1 | 100.0% | C | |
| GDPval-MM | 84.9% | 01 | 3 | 100.0% | C | |
| Terminal-Bench 2.0 | 82.7% | 01 | 49 | 100.0% | C | |
| GeneBench | 25.0% | 02 | 2 | 0.0% | C | |
| LiveBench | 80.7% | 02 | 38 | 97.3% | B | |
| MMMU-Pro | 83.2% | 02 | 66 | 98.5% | C | |
| CyberGym | 81.8% | 03 | 11 | 80.0% | C | |
| Graphwalks parents >128k | 58.5% | 03 | 7 | 66.7% | C | |
| Finance Agent | 60.0% | 04 | 8 | 57.1% | C | |
| MRCR v2 (8-needle) | 74.0% | 04 | 21 | 85.0% | C | |
| Finance Agent v2 | 51.8% | 05 | 26 | 84.0% | B | |
| FrontierCode 1.1 | 43.0% | 05 | 15 | 71.4% | B | |
| DeepSWE 1.1 | 67.0% | 06 | 20 | 73.7% | B | |
| FrontierSWE | 73.0% | 06 | 15 | 64.3% | B | |
| GPQA | 93.6% | 06 | 234 | 97.8% | C | |
| OfficeQA Pro | 54.1% | 06 | 7 | 16.7% | C | |
| Tau2 Telecom | 98.0% | 06 | 35 | 85.3% | C | |
| FrontierMath | 35.4% | 07 | 17 | 62.5% | C | |
| Graphwalks BFS >128k | 45.4% | 07 | 11 | 40.0% | C | |
| Legal Agent Benchmark | 2.1% | 07 | 13 | 50.0% | B | |
| Toolathlon | 55.6% | 07 | 31 | 80.0% | C | |
| OSWorld-Verified | 78.7% | 08 | 23 | 68.2% | C | |
| BrowseComp | 84.4% | 12 | 58 | 80.7% | C | |
| MCP Atlas | 75.3% | 13 | 31 | 60.0% | C | |
| Humanity's Last Exam | 52.2% | 16 | 93 | 83.7% | C | |
| SWE-Bench Pro | 58.6% | 16 | 45 | 65.9% | C |
GPT-5.5 Arena Results
Preference and agent-evaluation results for the default version.
| agent bash recovery steps | overall | 01 | 0.1 | N/A | 35K | |
| search factuality | overall | 01 | 1244.1 | 67,534 | N/A | |
| search | overall | 02 | 1240.3 | 67,573 | N/A | |
| search style control | overall | 02 | 1223.5 | 67,573 | N/A | |
| agent steerability | overall | 04 | 0.1 | N/A | 61.2K | |
| text factuality | overall | 05 | 1485.7 | 54,711 | N/A | |
| agent bash recovery steps | overall | 06 | 0.1 | N/A | 42.1K | |
| agent | overall | 07 | 0.1 | N/A | 1.9M | |
| agent tool hallucination | overall | 07 | 0.0 | N/A | 1.5M | |
| document | overall | 07 | 1484.7 | 16,816 | N/A | |
| agent bash recovery steps | overall | 08 | 0.1 | N/A | 36K | |
| agent praise complaint | overall | 08 | 0.1 | N/A | 11K | |
| agent steerability | overall | 08 | 0.1 | N/A | 42.1K | |
| agent tool hallucination | overall | 09 | 0.0 | N/A | 2M | |
| document | overall | 09 | 1480.3 | 17,286 | N/A | |
| text factuality | overall | 09 | 1481.9 | 56,002 | N/A | |
| agent | overall | 11 | 0.1 | N/A | 2.1M | |
| vision style control | overall | 12 | 1286.0 | 18,722 | N/A | |
| agent praise complaint | overall | 13 | 0.1 | N/A | 15.9K | |
| agent steerability | overall | 13 | 0.1 | N/A | 68.5K | |
| document style control | overall | 13 | 1477.8 | 16,816 | N/A | |
| agent | overall | 14 | 0.1 | N/A | 1.7M | |
| agent task outcome explicit | overall | 14 | 0.1 | N/A | 34.7K | |
| agent tool hallucination | overall | 14 | 0.0 | N/A | 1.8M | |
| document style control | overall | 15 | 1475.5 | 17,286 | N/A | |
| vision | overall | 15 | 1296.3 | 18,722 | N/A | |
| agent praise complaint | overall | 16 | 0.1 | N/A | 17.3K | |
| vision style control | overall | 16 | 1283.4 | 17,520 | N/A | |
| text style control | overall | 17 | 1481.4 | 54,790 | N/A | |
| agent task outcome explicit | overall | 19 | 0.0 | N/A | 52K | |
| vision | overall | 19 | 1293.9 | 17,520 | N/A | |
| agent task outcome explicit | overall | 20 | 0.0 | N/A | 55.1K | |
| text | overall | 20 | 1470.1 | 54,790 | N/A | |
| text style control | overall | 20 | 1476.6 | 56,079 | N/A | |
| text | overall | 23 | 1466.0 | 56,079 | N/A | |
| webdev | overall | 28 | 1508.5 | 12,558 | N/A | |
| webdev | overall | 33 | 1486.0 | 14,503 | N/A | |
| webdev | overall | 41 | 1457.9 | 12,613 | N/A |
GPT-5.5 Runtime Performance
Provider-specific output speed and catalog latency for GPT-5.5. Runtime does not affect the capability score.
| OpenAI | 70.342 tok/s | 113.442 s | 1.1M | 128K |
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
GPT-5.5 Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- $5 input, $30 output per 1M
- Official provider
- OpenAI
- Lowest third-party
- From $0.1875 input, $1.13 output per 1M via UnoRouter
- Tracked offerings
- 43
| Kenari | gpt-5-5 | global | N/A | N/A | 1.1M | |
| UnoRouter | gpt-5.5:free | global | N/A | N/A | 1.1M | |
| UnoRouter | gpt-5.5 | global | $0.1875 | $1.13 | 1.1M | |
| Xpersona | gpt-5.5 | global | $1.5 | $12 | 1.1M | |
| FrogBot | gpt-5-5 | global | $2.5 | $15 | 272K | |
| Xpersona | xpersona-gpt-5.5 | global | $3 | $18 | 1M | |
| Poe | openai/gpt-5.5 | global | $4.55 | $27.27 | 400K | |
| Impossibl | openai/gpt-5.5 | global | $5 | $30 | 1.1M | |
| CrossModel | openai/gpt-5.5 | global | $5 | $30 | 1.1M | |
| Vivgrid | gpt-5.5 | global | $5 | $30 | 1.1M | |
| NanoGPT | openai/gpt-5.5 | global | $5 | $30 | 1M | |
| FastRouter | openai/gpt-5.5 | global | $5 | $30 | 1.1M | |
| NEAR AI Cloud | openai/gpt-5.5 | global | $5 | $30 | 1.1M | |
| DaoXE | gpt-5.5 | global | $5 | $30 | 1.1M | |
| LLM Gateway | gpt-5.5 | global | $5 | $30 | 1.1M | |
| OpenCode Zen | gpt-5.5 | global | $5 | $30 | 1.1M | |
| GitHub Copilot | gpt-5.5 | global | $5 | $30 | 1.1M | |
| Neon | gpt-5-5 | global | $5 | $30 | 1.1M | |
| SAP AI Core | gpt-5.5 | global | $5 | $30 | 1.1M | |
| Pioneer | gpt-5.5 | global | $5 | $30 | 1.1M | |
| AI-ROUTER | gpt-5.5 | global | $5 | $30 | 1.1M | |
| ZenMux | openai/gpt-5.5 | global | $5 | $30 | 1.1M | |
| Abacus | gpt-5.5 | global | $5 | $30 | 1M | |
| Azure Cognitive Services | gpt-5.5 | global | $5 | $30 | 1.1M | |
| Azure | gpt-5.5 | global | $5 | $30 | 1.1M | |
| AIHubMix | gpt-5.5 | global | $5 | $30 | 1.1M | |
| Ofox | openai/gpt-5.5 | global | $5 | $30 | 1.1M | |
| OrcaRouter | openai/gpt-5.5 | global | $5 | $30 | 1.1M | |
| GMI Cloud | openai/gpt-5.5 | global | $5 | $30 | 1.1M | |
| FreeModel | gpt-5.5 | global | $5 | $30 | 1.1M | |
| Kilo Gateway | openai/gpt-5.5 | global | $5 | $30 | 1.1M | |
| Merge Gateway | openai/gpt-5.5 | global | $5 | $30 | 1.1M | |
| Vercel AI Gateway | openai/gpt-5.5 | global | $5 | $30 | 1M | |
| OpenRouter | openai/gpt-5.5 | global | $5 | $30 | 1.1M | |
| Cloudflare AI Gateway | openai/gpt-5.5 | global | $5 | $30 | 1.1M | |
| Perplexity Agent | openai/gpt-5.5 | global | $5 | $30 | 1.1M | |
| OpenAI | gpt-5.5 | global | $5 | $30 | 1.1M | |
| HPC-AI | openai/gpt-5.5 | global | $5 | $30 | 1.1M | |
| Databricks | databricks-gpt-5-5 | global | $5 | $30 | 1.1M | |
| Amazon Bedrock | openai.gpt-5.5 | global | $5.5 | $33 | 272K | |
| Venice AI | openai-gpt-55 | global | $6.25 | $37.5 | 1M |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
GPT-5.5 Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
| GPT-5.5 | 84.0 | N/A | 1.1M | 128K | No | Proprietary |
What is GPT-5.5?
Key information about GPT-5.5 and its available data.
5 is OpenAI's smartest model yet, designed for real work across agentic coding, computer use, knowledge work, and early scientific research. 4 per-token latency in real-world serving while reaching a much higher level of intelligence and using significantly fewer tokens to complete the same tasks.
0, OSWorld-Verified, GDPval, FrontierMath, and CyberGym.
Data as of 2026-08-11.
GPT-5.5 vs nearby models
Open a comparison with the three ranked models immediately above and below this model.
Models similar to GPT-5.5
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
OP
GPT-5.6-Terra
OpenAI
OP
GPT-5.4
OpenAI
OP
GPT-5.6-Luna
OpenAI
OP
GPT-5.5-Pro
OpenAI
OP
GPT-5.6-Sol
OpenAI
OP
GPT-5.2-Pro
OpenAI
FAQ
Common questions about GPT-5.5.
When was GPT-5.5 released?
GPT-5.5's default version was released on Apr 23, 2026.
How much does GPT-5.5 cost?
GPT-5.5's official API price is $5 per million input tokens and $30 per million output tokens via OpenAI. The lowest tracked third-party offer starts at $0.1875 input and $1.13 output via UnoRouter.
Who created GPT-5.5?
GPT-5.5 was created by OpenAI.
What is the context window for GPT-5.5?
The default version has a 1.1M token context window.
Is GPT-5.5 open weight?
No. The default version is not marked as having publicly available weights.
How many API providers offer GPT-5.5?
43 provider offerings are linked to the default version.
What models should I compare GPT-5.5 with?
Nearby ranked alternatives include GPT-5.6-Terra, Claude Opus 4.7, Claude Sonnet 5.