Loading...
Loading...
OpenAI has promoted GPT-5.5 Instant to be ChatGPT’s default model, replacing earlier Instant variants to deliver smarter, more accurate and personalized responses. The shift affects most users by changing the underlying model that generates answers, which could improve clarity, conciseness and perceived relevance without users changing settings. Details on benchmarks, pricing, rollout timing and tier availability remain sparse, but the move underscores OpenAI’s iterative approach to deployed models and highlights how default model choices materially shape user experience, latency and quality at scale.
OpenAI’s GPT-5.5 launch doubled per-token prices versus GPT-5.4 (input $2.50/M→$5.00/M; output $15/M→$30/M). TechScan AI replicated prior Opus analysis using an OpenRouter “switcher cohort” of users who moved from GPT-5.4 to 5.5 and measured token usage and billed costs. GPT-5.5 is less verbose on long prompts (≥10K tokens), cutting completions 19–34%, but is more verbose for 2K–10K prompts (+52%). Net effect: billed costs rose 49–92% across prompt-size buckets; costs for long prompts rose less due to shorter completions, while short-to-mid prompts saw larger percentage increases. The analysis uses OpenRouter token counts to normalize comparisons and excludes media and cancelled requests.
OpenAI’s GPT-5.5 launch doubled per-token prices versus GPT-5.4—input tokens from $2.50/M to $5.00/M and output tokens from $15/M to $30/M—and TechScan’s replication using OpenRouter logs finds user costs rose 49–92%. Analysts used a “switcher cohort” of users who moved from GPT-5.4 to 5.5 and normalized costs per million OpenRouter tokens. GPT-5.5 produces 19–34% fewer completion tokens for prompts over 10K tokens, partially offsetting higher prices; for shorter prompts under 10K tokens completions were similar or longer, driving larger cost jumps. The study covers April 21–28, 2026 windows and excludes media and cancelled requests—showing material increases in real-world billing depending on prompt sizes. This matters for teams budgeting LLM usage and for product decisions tied to model costs.
OpenAI : OpenAI launches GPT-5.5 Instant, which it says is smarter, with more accurate and personalized responses, replacing GPT-5.3 Instant as ChatGPT's default model — We're updating ChatGPT's default model, available to everyone, to be smarter and more accurate, with clearer, more concise answers that feel better tailored to you.
OpenAI has released GPT-5.5 Instant and is making it the new default model for ChatGPT, replacing GPT-3.5 Instant. The change means most ChatGPT users will now be served responses from GPT-5.5 Instant by default rather than the older GPT-3.5 Instant model. The article provides no additional details on performance, pricing, rollout timing, or availability across tiers, and does not include benchmarks or technical specifications. Even with limited information, the update matters because the default model choice affects response quality, latency, and user experience at scale, and signals OpenAI’s ongoing iteration of its deployed language models within ChatGPT.