Gemini 2.5 Flash: clarified the exact input limit and added the documented maximum output length
Google updates technical specifications for Gemini 2.5 Flash to ensure developer reliability.
Clarified the exact input limit and added the documented maximum output length — verified by the OnAirToday AI Progress Tracker against Google DeepMind's published documentation.
OnAirToday
Newsroom

The OnAirToday AI Progress Tracker recorded a verified change to Google DeepMind's Gemini 2.5 Flash (version gemini-2.5-flash): Clarified the exact input limit and added the documented maximum output length.
Gemini 2.5 Flash is Google's multimodal model for balancing reasoning quality, latency and cost, with configurable thinking and text output. Published standard Gemini API pricing is $0.30 per million text, image or video input tokens and $2.50 per million output tokens, including thinking tokens; audio input is priced separately.
What changed in this review:
- Clarified that standard output pricing includes thinking tokens and audio input has separate pricing.
- Added published GPQA Diamond, AIME 2025 and MMMU results.
- Excluded unconfirmed Terminal-Bench and FrontierMath figures.
- Added two overlapping reports of Gemini-brand phishing, without attributing an underlying vulnerability to Gemini 2.5 Flash.
- Capability score 12 → 78
- Flaws and limitations updated
Anyone building on Gemini 2.5 Flash should re-check the model's risk dashboard before shipping, since flaws and reported incidents are what moved in this review.
The full profile — capabilities, benchmark citations, known flaws and risk guidance — is on the tracker at /ai-models/gemini-2-5-flash.
References:
- https://ai.google.dev/gemini-api/docs/models#gemini-2.5-flash
- https://ai.google.dev/gemini-api/docs/pricing#gemini-2.5-flash
- https://blog.google/technology/google-deepmind/gemini-2-5-model-family-expands/
- https://arxiv.org/abs/2507.06261
Generated insight by OnAirToday AI
Explicitly documenting input and output limits provides essential guardrails for developers building scalable, enterprise-grade AI applications. This transparency reduces integration uncertainty and helps engineers optimize prompt architectures within hard computational constraints.
Get the day's AI news in one short email. Free.
Comments
Keep comments relevant and respectful. No links, advertising, self-promotion, or spam.