gpt-5-5-instant
Overview
GPT-5.5 Instant is an OpenAI model released on May 5, 2026 as the successor to GPT-5.3 Instant. It shares the same underlying architecture as GPT-5.5 Thinking and Pro, a natively omnimodal model that processes text, images, audio, and video end to end, but is optimized for low latency and everyday conversational use.
Key Features
- Reduced Hallucinations: 52.5% fewer hallucinated claims than GPT-5.3 Instant on high-stakes prompts such as medicine, law, and finance, and 37.3% fewer inaccurate claims on user-flagged factual errors.
- Natively Omnimodal: Processes text, images, audio, and video in a single unified architecture rather than combining separate models. Improvements include better image understanding and stronger STEM question answering.
- Concise Output: Uses roughly 30.2% fewer words and 29.2% fewer lines than GPT-5.3 Instant to convey the same information, with tighter formatting and less unnecessary verbosity.
- Personalized Memory: Can reference past conversations, files, and Gmail for more personalized responses for eligible Plus and Pro users.
Best Use Cases
- Everyday Knowledge Work: Optimized for information-seeking questions, how-tos, technical writing, and translation with a warm conversational tone and low latency.
- Multimodal Analysis: Strong image and document understanding makes it well-suited for analyzing uploads, screenshots, charts, and other visual content.
- High-Stakes Factual Q&A: The reduction in hallucinations makes it more reliable for queries in medicine, law, and finance compared to prior Instant-tier models.
Capabilities and Limitations
| Capability | Description |
|---|---|
| Reasoning | AIME 2025: 81.2% versus 65.4% for GPT-5.3 Instant. Shares architecture with GPT-5.5 Thinking |
| Coding | Capable for everyday coding tasks; GPT-5.5 Thinking is recommended for complex agentic coding |
| Multimodal | Text, image, audio, and video input in a natively omnimodal architecture |
| Response Speed | Low-latency design; matches GPT-5.4 per-token latency despite higher capability |
| Context Window | 1M tokens (922K input + 128K output). Long-context surcharge above 272K input tokens |
| Max Output | 128K tokens |
| Tool Use | Web search, file analysis, and auto-switching across tools |
| Multilingual | Improved translation quality with broad multilingual support |
Known Limitations
- Instant tier trades reasoning depth for speed, so complex multi-step reasoning and agentic workflows are better served by GPT-5.5 Thinking or Pro.
- MMMU-Pro multimodal score (76) and AIME math score (81.2) are lower than the full GPT-5.5 Thinking model.
- The rapid release cadence means prompts and custom GPT workflows optimized for this model may need periodic retuning.
- Knowledge cutoff is December 2025. Web search can help with more recent information.
Standard Pricing
The token prices below are shown in USD per 1 million tokens; web search is billed in USD per use.
| Model | Input (USD / 1M Tokens) | Cache Write (USD / 1M Tokens) | Cache Read (USD / 1M Tokens) | Output (USD / 1M Tokens) | Web Search (USD / use) | Billing Notes |
|---|---|---|---|---|---|---|
| GPT-5.5 Instant | $5.00 | $5.00 | $0.50 | $30.00 | $0.01 | Long-context (>272K input tokens): Input 2x, Output 1.5x |
Credits settlement: B.AI converts charges at 1 USD = 1,000,000 Credits and deducts Credits from the account balance.
Prices shown in the documentation are B.AI standard reference prices for base billing purposes. B.AI may provide lower actual usage costs through top-up bonuses and account benefits. Specific prices, bonus Credits, and account benefits are subject to the platform display and final billing records.