Apply your own routing policies, reduce token costs automatically, and see every routing decision in real time with Merge Gateway.
What Amazon Nova 2 Lite costs to run
| Vendor | Input / 1M tokens | Output / 1M tokens | Zero data retention |
| --- | ---: | ---: | --- |
| Amazon Bedrock | $0.3000 | $2.50 | Yes |
Test Amazon Nova 2 Lite
with Gateway’s Simulator
See a prompt's output, token spend, latency, and more with Amazon Nova 2 Lite.
Route requests to Amazon Nova 2 Lite in minutes
To get started in seconds, add our Gateway Implementation skill to your project, or pick your preferred SDK below. Check out our other quick start skills here.
Install the Merge Gateway SDK
Python
1$ pip install merge-gateway-sdkSend a request
Python
1from merge_gateway import MergeGateway
2
3client = MergeGateway(api_key="YOUR_API_KEY")
4
5response = client.responses.create(
6 model="openai/gpt-5.2",
7 input=[
8 {"type": "message", "role": "system", "content": "You are a helpful programming tutor. Explain the concepts clearly with practical examples."},
9 {"type": "message", "role": "user", "content": "Explain the concept of recursion in programming with a simple set of examples."},
10 ],
11)
12
13print(response.output[0].content[0].text)Try a diffrent model
Swap the model string to route to a different provider. No other code changes needed.
Anthropic
1response = client.responses.create(
2 model="anthropic/claude-sonnet-4-20250514",
3 input=[
4 {"type": "message", "role": "system", "content": "You are a helpful programming tutor. Explain the concepts clearly with practical examples."},
5 {"type": "message", "role": "user", "content": "Explain the concept of recursion in programming with a simple set of examples."},
6 ],
7)Point to Gateway
Python
1from openai import OpenAI
2
3client = OpenAI(
4 api_key="YOUR_API_KEY",
5 base_url="https://api-gateway.merge.dev/v1/openai",
6)Send a request
Use the standard chat.completions.create method. No provider prefix needed on the model name.
Python
1response = client.chat.completions.create(
2 model="gpt-5.2",
3 messages=[
4 {"role": "system", "content": "You are a helpful programming tutor. Explain the concepts clearly with practical examples."},
5 {"role": "user", "content": "Explain the concept of recursion in programming with a simple set of examples."},
6 ],
7)
8
9print(response.choices[0].message.content)Install packages
1npm install merge-gateway-ai-sdk-provider aiCreate the provider
TypeScript
1import { createMergeGateway } from "merge-gateway-ai-sdk-provider";
2
3const gateway = createMergeGateway({
4 apiKey: "YOUR_API_KEY",
5});Send a request
Use generateText to send a request. Model names use the provider/model format.
TypeScript
1import { generateText } from "ai";
2
3const { text } = await generateText({
4 model: gateway("openai/gpt-4o"),
5 prompt: "Explain the concept of recursion in programming with a simple set of examples.",
6});
7
8console.log(text);If you already have @ai-sdk/openai installed, point it at Gateway with a base URL change:
TypeScript
1import { createOpenAI } from "@ai-sdk/openai";
2
3const gateway = createOpenAI({
4 apiKey: "YOUR_API_KEY",
5 baseURL: "https://api-gateway.merge.dev/v1/ai-sdk",
6});
7
8// All generateText/streamText calls work unchangedInstall the Merge Gateway SDK
Anthropic SDK
1from anthropic import Anthropic
2
3client = Anthropic(
4 api_key="YOUR_API_KEY",
5 base_url="https://api-gateway.merge.dev/v1/anthropic",
6)
7
8message = client.messages.create(
9 model="claude-sonnet-4-20250514",
10 max_tokens=1024,
11 messages=[
12 {"role": "user", "content": "Explain the concept of recursion in programming with a simple set of examples."},
13 ],
14)
15
16print(message.content[0].text)Explore other models available in Merge Gateway
Amazon Nova 2 Lite FAQ
For anyone with more questions about Amazon Nova 2 Lite, we've covered a few more below. The details here reflect what was known in 06/2026 and are subject to change.
Heading
What other models does AWS offer?
Amazon's Nova family on Amazon Bedrock spans text, multimodal, and media-generation models across several price and capability tiers. Here are some other models AWS supports:
- Amazon Nova 2 Pro: The most intelligent model in the Nova 2 line, built for highly complex, multistep tasks, and available in preview (as of 06/24/2026) for teams that need more reasoning depth than Nova 2 Lite provides
- Amazon Nova Premier: The most capable model in the first-generation Nova lineup, aimed at complex reasoning and well suited to teaching or distilling smaller, cheaper custom models
- Amazon Nova Pro: A balanced first-generation multimodal model that trades some of Premier's capability for lower cost and faster responses on everyday production workloads
- Amazon Nova Micro: A text-only model optimized for the lowest cost and latency, a fit for high-volume classification, routing, and simple generation where multimodal input isn't needed
- Amazon Nova Canvas: An image-generation model for producing and editing visuals from text prompts
- Amazon Nova Sonic: A speech model for real-time, low-latency voice interactions, useful for conversational and voice-agent applications
How does Amazon Nova 2 Lite differ from AWS's other models?
Nova 2 Lite sits in the middle of the Nova 2 line: a reasoning-capable model tuned for speed and cost on everyday workloads, below Nova 2 Pro and above the text-only first-generation Micro.
- Extended thinking: Unlike text-only Nova Micro, Nova 2 Lite supports step-by-step reasoning and task decomposition with three thinking intensity levels (low, medium, high), so you can dial reasoning effort up or down per request
- Context window: A 1M-token context window, large enough for long documents and multi-file inputs, matching Nova 2 Pro and exceeding most first-generation Nova tiers
- Modalities: Accepts text, image, video, and document input, with built-in web grounding and a code interpreter, where Nova Micro is text-only
- Pricing: Positioned as the cost-effective Nova 2 option, at $0.30 per 1M input tokens and $2.50 per 1M output tokens (as of 06/24/2026), well below Nova 2 Pro
- Use case fit: AWS highlights instruction following, math, video understanding with temporal reasoning, and practical code generation, versus Nova 2 Pro's focus on the hardest multistep problems
Nova 2 Lite is the right pick when you want agentic, reasoning-capable behavior and multimodal input at a low per-request cost, and you don't need the full intelligence of Nova 2 Pro.
What models should I consider using alongside Amazon Nova 2 Lite?
No single model is optimal for every task. Here are models worth pairing with Amazon Nova 2 Lite depending on what your product needs:
- Claude Opus 4.8 for the hardest reasoning and large-codebase work, where you'd route the small fraction of requests that need maximum depth and accept the higher cost for those alone
- GPT-5.2 for complex agentic tool-use and tasks needing strong general intelligence, when Nova 2 Lite's reasoning isn't sufficient and you want a frontier alternative outside the AWS ecosystem
- Gemini 2.5 Flash for ultra-high-volume, latency-sensitive calls where you want a second low-cost multimodal option and provider diversity away from Bedrock
- DeepSeek V4 for cost-efficient reasoning on text-heavy tasks, as a cheaper fallback when a request needs deliberate reasoning but not Nova 2 Pro or Claude Opus
- Amazon Nova Micro for the simplest, highest-volume text tasks (classification, routing, extraction), where Nova 2 Lite's reasoning and multimodal capabilities are overkill and Micro's lower cost wins
What are the challenges of using Amazon Nova 2 Lite in my product?
Like any production LLM, Amazon Nova 2 Lite comes with tradeoffs worth planning for:
- Output-token cost on reasoning: Output tokens are priced well above input tokens, and extended thinking generates more of them, so reasoning-heavy requests cost more than the low input rate suggests
- Reasoning latency: Higher thinking intensity adds tokens and time to first usable answer, so latency-sensitive paths may need the low-intensity setting or a lighter model
- Provider dependency: Running solely on Nova 2 Lite ties you to AWS Bedrock, so a Bedrock outage, a regional capacity limit, or a model version change becomes your outage with no automatic fallback
- Bedrock availability and access: Access runs through Amazon Bedrock, which means region availability, model-access enablement, and Bedrock quotas all gate your usage, and Nova 2 Pro was still in preview as of 06/24/2026
- Cost at scale: As request volume grows, token costs compound quickly, and without active routing and budgets a reasoning model's output spend can climb faster than expected
Why should I use Merge Gateway to route LLM requests with Amazon Nova 2 Lite and every other model?
Using Amazon Nova 2 Lite through Merge Gateway gives you access to the model itself and the infrastructure layer around it:
- One API, every provider: Reach Amazon Nova 2 Lite and every other major LLM through a single endpoint and API key. Switch models by swapping the model string, with no application code changes
- Intelligent routing and automatic failover: Merge routes around AWS Bedrock outages automatically, and routing policies based on cost, latency, or quality can cut spend by 40 to 60 percent without touching your code
- Cost governance: Set hard or soft project budgets so Nova 2 Lite's output-token spend stays within plan, with every request attributed to a model, project, and tag in one billing dashboard across providers
- Build Your Own Router: Define what "best" means for your traffic using curated ML benchmarks or your own eval scores. The router scores each available model against your weights and picks the winner per request, with a plain-language explanation of every decision
- Security and compliance controls: Apply DLP rules and prompt injection protection before any request reaches AWS, and enforce per-project model and region policies without building that logic into your application
How can I start routing requests to Amazon Nova 2 Lite via Merge Gateway?
Getting Amazon Nova 2 Lite running through Merge Gateway takes a few minutes:
1. Create an account and get your API key from the dashboard.
2. Install the Merge Gateway SDK: run pip install merge-gateway-sdk (Python) or npm install merge-gateway-sdk (Node). Alternatively, if you're already using the OpenAI SDK, set base_url = "https://api-gateway.merge.dev/v1/openai" and your existing code works as-is.
3. Make your first request using the provider/model format. For Amazon Nova 2 Lite, the model string is amazon/nova-2-lite (confirm the exact slug in the Merge Gateway model catalog, since Bedrock IDs use the amazon.nova-2-lite-v1:0 form). Swap the model string to route to any other provider without changing anything else.
4. Configure a routing policy in the dashboard to set failover behavior, cost limits, and optimization strategy. Your first policy can be as simple as naming Amazon Nova 2 Lite as primary with one fallback.
Full setup instructions and SDK references are in the Merge Gateway docs.
Try Amazon Nova 2 Lite through Merge Gateway
Route, observe, and control AI requests across providers from one API.
