llmrelay
// gemini

Gemini 3.5 Flash

Fast and cheap, with the full 1M context window.

Gemini 3.5 Flash is Google’s fast, cheap model with native multimodal input - text, images, video, audio all in one API call. It carries the full 1M-token context window while costing less per token than GPT-5.6 Luna or Claude Haiku, making it the go-to for high-volume multimodal work.

The "Flash" designation means speed: Google rates it as optimised for low latency, so it can sit in a request path without users feeling the delay. That makes it the right model for production endpoints, RAG pipelines, and batch jobs where you are running thousands of calls.

We sell it at half Google’s list price - $0.75 per million input tokens against $1.50. At that rate it is the cheapest model we sell with a 1M context window, which opens up long-document work that used to require flagship pricing.

Official list
llmrelay
Input / M tokens
$1.50
$0.75
Output / M tokens
$9.00
$4.50
Context window
—
1,048,576 tokens
Max output
—
65,536 tokens

What it costs you per month

Real-world budget scenarios. Numbers are simple sums — official list price vs llmrelay's 50% off tier.

Usage scenario
Official
llmrelay
Light coding (1M in / 200K out per month)
$3.30
$1.65save $1.65
Heavy Cursor / Cline user (50M in / 5M out per month)
$120.00
$60.00save $60.00
Production RAG (500M in / 20M out per month)
$930.00
$465.00save $465.00

What it's good at

Best for

Pick Gemini 3.5 Flash when

  • +Multimodal work on a budget. Flash natively handles images, video, and audio without needing separate preprocessing pipelines, and it is cheap enough to run at volume.
  • +High-volume summarization or RAG. The 1M context window at $0.75/M input makes Flash the most cost-effective way to put entire documents in front of a model.
  • +Batch tagging or classification where you need long context. Flash costs less than Haiku or Luna while carrying 5× the context window.
  • +You are exploring multimodal LLM work and want to prototype cheaply before committing to a more expensive model.

Choose something else when

  • !Complex reasoning or multi-step agent work. Flash is optimised for speed and cost, not depth. Use Claude Opus, Claude Sonnet, or GPT-5.6 Sol/Terra for reasoning tasks.
  • !Work where you need the strongest accuracy. Gemini 3.1 Pro is Google’s flagship and costs more for a reason. If correctness is load-bearing, do not optimise on price alone.
  • !Paying a Claude flagship for ordinary copy. Fable is not a writing SKU; for hard Claude work the current id is Fable 5.1.
  • !Function calling or structured output as a primary workflow. Flash supports it, but OpenAI’s and Anthropic’s tool-use implementations are more mature.

Questions people ask about Gemini 3.5 Flash

How much does Gemini 3.5 Flash cost?

Google’s list price is $1.50 per million input tokens and $9.00 per million output tokens. On llmrelay it is $0.75 and $4.50 - exactly half list. Billing is prepaid per token from one credit pool, with no subscription.

Can Gemini 3.5 Flash handle images and video?

Yes. Gemini models natively support multimodal input - text, images, video, and audio all in one API call. You do not need separate preprocessing pipelines or vision-specific endpoints.

Should I use Gemini 3.5 Flash or Claude Haiku 4.5?

Flash if you need long context (1M vs 200k) or native multimodal input. Haiku if you are staying in the Claude ecosystem and do not need those features. Flash costs $0.75/M input against Haiku’s $0.50, so Haiku is cheaper if context and multimodal do not matter.

What is the context window on Gemini 3.5 Flash?

1,048,576 tokens input (1M), with up to 65,536 output tokens. That is the same input window as Gemini 3.1 Pro and larger than Claude Haiku or most small models, making Flash unusually capable for long-document work at its price point.

Try Gemini 3.5 Flash at half the price

Free to create an account, no subscription, $10 minimum top-up — enough to run this model against a real task and judge quality yourself.

Get API key →

Compare Gemini 3.5 Flash against the alternatives

The comparisons this model appears in, the models nearest it on price, and the full rate card.