AI News · Models ·
OpenAI releases GPT-5.4 mini and nano for faster, lower-cost AI
OpenAI is releasing GPT-5.4 mini and nano, two models designed for speed and affordability. The mini model is reported to be more than twice as fast as GPT-5 mini. It also performs close to the full-size model on key benchmarks, according to the report.
Key points
- OpenAI is releasing GPT-5.4 mini and nano for speed and affordability.
- NewsBytes reported mini is more than twice as fast as GPT-5 mini.
- Mini costs $0.75 per million input tokens and $4.50 per million output tokens.
- Nano costs $0.20 per million input tokens and $1.25 per million output tokens.
- Mini reportedly approaches full-size benchmark performance, but benchmark names, scores and test conditions were not reported.
What happened: OpenAI is releasing GPT-5.4 mini and GPT-5.4 nano, two AI models designed for faster responses and lower costs, NewsBytes reported. The release targets developers and businesses looking for affordable AI tools, including applications where response time matters. According to the report, OpenAI said its aim is to make powerful AI more widely accessible without sacrificing quality or speed. The reported performance details focus on mini rather than nano.
The details: GPT-5.4 mini costs $0.75 per million input tokens and $4.50 per million output tokens, according to NewsBytes. GPT-5.4 nano is priced at $0.20 per million input tokens and $1.25 per million output tokens. These are separate rates for input and output, not a single price covering both. Nano has the lower listed rate in each category. Actual costs for a business workload were not reported, so the listed prices alone do not establish what a particular application will cost to run.
Background: NewsBytes reported that GPT-5.4 mini is more than twice as fast as GPT-5 mini and performs almost on par with the full-size model on key benchmarks. Those comparisons address two distinct questions for buyers: how quickly a model responds and how well it performs. The report did not name the benchmarks, give scores or describe the conditions used to measure speed. It also did not report comparable speed or benchmark results for nano, leaving less detail about the cheaper model's performance.
Who it affects: Coding assistants and applications where quick replies matter are among the uses highlighted by NewsBytes. For teams choosing models for those applications, mini's reported speed improvement is a reason to consider testing it, rather than proof that it will meet every requirement. The near-full-size benchmark performance is also worth examining against representative tasks. The report does not establish whether that result carries over to a particular company's coding work or other business needs.
What to watch: Teams evaluating either model should verify response times, answer quality and actual costs on their own tasks before making a deployment decision. The available comparison is more specific for mini; nano's lower prices are clear, but its relative quality and speed were not reported. Access details and rollout timing were also not reported.
Our take
Smaller models are worth testing for latency-sensitive applications, but teams should verify quality and actual costs on representative tasks.