Gemini 3.7 Flash
Gemini 3.7 Flash is Google's flagship high-efficiency workhorse model engineered for autonomous agentic workflows and native multimodal reasoning. Built on enhanced core algorithmic foundations, it delivers ultra-low latency inference alongside robust multi-step tool execution, adaptive task planning, and end-to-end software engineering. Operating at $0.75 per million input tokens, Gemini 3.7 Flash powers enterprise agent orchestration, low-latency conversational interactive applications, real-time code generation, and high-volume data processing pipelines.
Related Models
Gemini 3.8 Flash
Gemini 3.8 Flash is Google's most intelligent Flash-tier model, engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows. Featuring a 1M-token context window and up to 64k output tokens, it introduces tunable reasoning levels (Low, Medium, High) to balance output latency and execution depth. Built for resilient multi-step planning, deterministic tool orchestration, and complex multi-file code refactoring, Gemini 3.8 Flash minimizes iterative execution errors—powering autonomous AI agents, enterprise data pipelines, technical software refactoring, and high-concurrency API deployment.
Gemini Omni Flash Text-to-Video
Gemini Omni Flash Text-to-Video is Google DeepMind's flagship unified multimodal model under the Gemini Omni series. Integrating Gemini's advanced reasoning with Veo's video generation capability, it enables native "any-to-any" generation, producing high-quality 24 FPS videos with synchronized audio directly from text prompts. It supports 16:9 and 9:16 aspect ratios, 3–10s video duration, and up to 1080P/4K supersampled output. Featuring multi-turn conversational video editing with context retention and built-in SynthID digital watermarking, it is ideal for short-form video creation, commercial ads, film VFX, and multimodal Agent workflows.
Gemini Omni Flash Image-to-Video
Gemini Omni Flash Image-to-Video is a next-generation multimodal video generation model developed by Google DeepMind. Built on a native Omni architecture, it accurately parses text prompts and input image semantics to produce cinematic 24 FPS dynamic videos. Supporting 16:9 and 9:16 aspect ratios, it generates 3–10 second fluid clips per run (featuring native support for up to 4K super-sampled upscaling, with 720P currently available on select platform endpoints). With exceptional subject consistency, physical simulation, and camera control, it excels in short-form drama, commercial advertising, film VFX, and social media animation.
Gemini 3.6 Flash
Gemini 3.6 Flash is Google's next-generation lightweight workhorse model released in July 2026. Built for agentic workflows, complex coding, and multimodal tasks, it supports a 1M token input context window and a 64K token output limit. Compared to 3.5 Flash, it reduces output token consumption by ~17% with streamlined reasoning steps and tool calls, significantly lowering overall cost and latency for agent execution. Natively handling text, image, video, audio, and PDF inputs, it excels at computer use and multi-tool orchestration.
Read Me
Gemini 3.7 Flash
Gemini 3.7 Flash is a multimodal large language model released by Google DeepMind on August 13, 2026. Based on Gemini 3.6 Flash with algorithmic improvements to its core reasoning foundation, it supports customizable thinking configurations to control the mix of quality, cost, and latency. The model accepts text, image, audio, video, and PDF input with text output, a 1 million token context window, 64K max output, and a knowledge cutoff of March 2026. It ranks first on GPQA Diamond, Code Arena WebDev, and GDM-MRCR v2, with FrontierCode 1.1 at 43.6% (up from 3.6 Flash's 34.4%) and DeepSWE v1.1 at 65.3%.
The iCreat platform offers this model via the endpoint https://api.icreat.ai/llm/openai/v1/chat/completions (OpenAI Chat Completions compatible), at $0.75 per million input tokens and $3.75 per million output tokens — identical to the official pricing.
Model Positioning
Gemini 3.7 Flash is positioned as Google DeepMind's multimodal workhorse model, suited for coding, agentic workflows, and enterprise tasks.
Based on Gemini 3.6 Flash with algorithmic improvements, the model delivers comprehensive upgrades across reasoning, coding, agentic tool use, multimodal capabilities, and long-context performance. Customizable thinking levels (low/medium/high) let developers balance quality, cost, and latency, defaulting to medium.
On the iCreat platform, the model is served through an OpenAI Chat Completions compatible endpoint, and existing OpenAI SDKs work directly.
Core Capabilities
Omni-Modal Input
Supports text, image, audio, video, and PDF input with text output (64K max tokens), covering document understanding, chart analysis, video comprehension, and audio processing multimodal scenarios.
1 Million Token Ultra-Long Context
Supports 1,048,576 token context window with 97.0% accuracy on 128K long-context retrieval (GDM-MRCR v2 8-needle), enabling processing of large codebases and long documents in a single call.
Customizable Thinking Configuration
Supports three thinking levels (low/medium/high, default medium), allowing developers to flexibly balance reasoning depth against cost and latency. The model invests more compute in multi-step planning and tool calls, reducing retries in agentic workflows.
Coding and Agent Capabilities
FrontierCode 1.1 at 43.6% (first), Code Arena WebDev Elo 1588 (first), DeepSWE v1.1 at 65.3%, Terminal-Bench 2.1 at 85.8% — leading in coding and agentic terminal tasks.
Streaming Output and Function Calling
Supports streaming output, function calling, structured output, code execution, prompt caching, and batch processing across the Chat Completions endpoint.
Pricing
| Token Type | Unit Price | Unit |
|---|---|---|
| Input | $0.75 | per million tokens |
| Output | $3.75 | per million tokens |
| Cached Read | $0.07 | per million tokens |
| Cache Write | $0.04 | per million tokens |
Note: The above are iCreat platform prices, identical to Google's official pricing (input $0.75, output $3.75, cache read $0.075). Official promotional pricing expires December 31, 2026; starting January 1, 2027, $1.50/$7.50 will apply. The iCreat platform price shall prevail.
Application Scenarios
- Agentic workflows and multi-step tool calling
- Coding tasks and software engineering (multi-file development, debugging, refactoring)
- Long-document analysis and large-scale codebase understanding
- Multimodal document processing (charts, PDFs, video, audio)
- Enterprise workflow automation
Model Comparison
Comparison Table 1: Gemini 3.7 Flash vs 3.6 Flash
| Feature | Gemini 3.7 Flash | Gemini 3.6 Flash |
|---|---|---|
| Positioning | 3.7 algorithmic improvement | 3.6 base version |
| Release Date | 2026-08-13 | 2026 |
| Context Window | 1M tokens | 1M tokens |
| Max Output | 64K tokens | 64K tokens |
| Knowledge Cutoff | 2026-03 | Earlier |
| FrontierCode 1.1 | 43.6% | 34.4% |
| DeepSWE v1.1 | 65.3% | 48.6% |
| OSWorld 2.0 | 47.9% | 33.8% |
| Input Price | $0.75/M tokens | $0.75/M tokens |
| Output Price | $3.75/M tokens | $3.75/M tokens |
Note: 3.7 Flash improves across coding, agents, and long-context at the same price as 3.6 Flash.
Comparison Table 2: Same-Tier Multimodal Models
| Feature | Gemini 3.7 Flash | Claude Sonnet 5 | GPT-5.6 Terra |
|---|---|---|---|
| Positioning | Multimodal workhorse | Enterprise coding model | General flagship |
| Input Price | $0.75/M tokens | $2.00/M tokens | $2.00/M tokens |
| Output Price | $3.75/M tokens | $10.00/M tokens | $12.00/M tokens |
| Context Window | 1M tokens | 1M tokens | 1M tokens |
| Input Modalities | Text+Image+Audio+Video+PDF | Text+Image | Text+Image |
| AA Intelligence Index | 56 | 55 | 57 |
| FrontierCode 1.1 | 43.6% | 42.7% | 41.3% |
Note: Gemini 3.7 Flash is the lowest-priced of the three with the top coding score and the broadest input modality support (only one supporting audio+video+PDF).
Why Choose Gemini 3.7 Flash?
- Omni-modal input (text+image+audio+video+PDF), covering the broadest multimodal scenarios
- 1 million token ultra-long context with 97.0% retrieval accuracy at 128K
- First place on both FrontierCode 1.1 and Code Arena WebDev, leading coding capability
- Customizable thinking levels (low/medium/high) for flexible quality-cost balance
- Lowest price among comparable models ($0.75/$3.75), far below Sonnet 5 and GPT-5.6
- OpenAI Chat Completions compatible endpoint — existing SDKs work directly
Specifications
| Field | Value |
|---|---|
| Model Name | Gemini 3.7 Flash |
| Developer | Google DeepMind |
| Model ID | gemini-3.7-flash |
| Endpoint | https://api.icreat.ai/llm/openai/v1/chat/completions |
| SDK base_url | https://api.icreat.ai/llm/openai/v1 |
| Release Date | 2026-08-13 |
| Model Type | Multimodal LLM |
| Authentication | Authorization: Bearer |
| Context Window | 1,048,576 tokens (1M) |
| Max Output | 65,536 tokens (64K) |
| Knowledge Cutoff | 2026-03 |
| Input Modalities | Text, Image, Audio, Video, PDF |
| Output Modalities | Text |
| Thinking Level | low / medium (default) / high |
| Streaming | Supported |
| Function Calling | Supported |
| Structured Output | Supported |
| Code Execution | Supported |
| Prompt Caching | Supported |
| Batch Processing | Supported |
| Billing Unit | Per million tokens |
Architecture
The iCreat platform's Gemini 3.7 Flash is served through an OpenAI Chat Completions compatible endpoint, allowing users to call it with any OpenAI API-compatible SDK. Requests must include the API Key in the Authorization header. The request body contains the model field (gemini-3.7-flash), a messages array (supporting multimodal content: text, image_url, video_url, input_audio), and a modalities field (specifying output modality). The model is based on Gemini 3.6 Flash with algorithmic improvements, supporting customizable thinking levels (low/medium/high) and investing more compute in multi-step planning and tool calls. The model supports streaming (stream: true), function calling, structured output, code execution, and prompt caching.
Notes
- The model supports omni-modal input (text+image+audio+video+PDF) but output is text only
- Max output is 64K tokens, smaller than Claude Opus 5 (128K) and Fable 5.1 (128K)
- Thinking level defaults to medium; adjust to low or high based on task complexity
- Official promotional pricing ($0.75/$3.75) expires 2026-12-31; starting 2027-01-01, $1.50/$7.50 applies
- API Key should be passed via environment variables; avoid hardcoding it in source code
- Please safeguard your API Key and avoid hardcoding it in client-side code or public repositories
Frequently Asked Questions
What is the difference between Gemini 3.7 Flash and 3.6 Flash?
3.7 Flash is based on 3.6 Flash with algorithmic improvements, delivering comprehensive upgrades in coding (FrontierCode 43.6% vs 34.4%), agents (DeepSWE 65.3% vs 48.6%), and computer use (OSWorld 47.9% vs 33.8%), at the same price as 3.6 Flash.
What input modalities are supported?
Supports text, image, audio, video, and PDF — five input modalities. Output is text only (64K max tokens). In the content array, specify input type via the type field: text, image_url, video_url, input_audio.
How do I control the thinking level?
Three levels: low, medium (default), and high. Lower levels reduce cost and latency; higher levels improve reasoning depth. Use medium for everyday tasks and high for complex reasoning.
How large is the context window?
Supports 1 million tokens (1,048,576 tokens) context window with 64K (65,536 tokens) max output. Achieves 97.0% accuracy on 128K long-context retrieval (GDM-MRCR v2 8-needle).
Is streaming output supported?
Yes. Pass "stream": true in the request body to enable streaming. The model will return generated content token by token.
Is function calling supported?
Yes. The model is compatible with the OpenAI Function Calling protocol, and also supports structured output (JSON Schema), code execution, prompt caching, and batch processing. Native web search is not supported (use external search tools via function calling).