Last refreshed 2026-11-17. Next refresh: weekly.
Why use d1.2-ultra on Vercel AI Gateway?
Vercel AI Gateway offers d1.2-ultra with pay-as-you-go pricing at $0.04/1M input tokens. Vercel AI Gateway is a unified AI proxy providing a single OpenAI-compatible API endpoint to 275+ models from 25+ providers including Anthropic, OpenAI, Google, Meta, Mistral, DeepSeek, xAI, Alibaba, Amazon, ByteDance, Cohere, MiniMax, MoonshotAI, KwaiPilot, Black Forest Labs, Recraft, Voyage AI, NVIDIA, and more.
Compare d1.2-ultra across 3 providers to find the best fit for your use caseSetup recipe
Python + curlpip install openaiexport AI_GATEWAY_API_KEY=...import os
from openai import OpenAI
client = OpenAI(
api_key=os.environ["AI_GATEWAY_API_KEY"],liquid/d1.2-ultraRequest example
import os
from openai import OpenAI
client = OpenAI(
api_key=os.environ["AI_GATEWAY_API_KEY"],
base_url="https://ai-gateway.vercel.sh/v1"
)
response = client.chat.completions.create(
model="liquid/d1.2-ultra",
messages=[{"role": "user", "content": "Hello"}]
)
print(response.choices[0].message.content)Gotchas
- Use provider model ID "liquid/d1.2-ultra", not the LLMReference slug "d1.2-ultra".
- creator/model-name e.g. kwaipilot/kat-coder-pro-v2
- The examples expect AI_GATEWAY_API_KEY; rename it only if your application config maps the new variable.
Compare d1.2-ultra Across Providers
| Provider | Input (per 1M) | Output (per 1M) |
|---|---|---|
| Liquid AI | $0.04 | Free |
| OpenRouter | $0.04 | Free |
| Vercel AI Gateway | $0.04 | Free |
Pricing
| Type | Price (per 1M) |
|---|---|
| Input tokens | $0.04 |
| Output tokens | Free |
Capabilities
About d1.2-ultra
d1.2-ultra is the larger, higher-accuracy version of Liquid AI's d1.2 decision models, released November 16, 2026 on the Liquid API as the update to d1-ultra. Like earlier d1 models it reads text and images, plus typed questions, and returns calibrated answers in a single forward pass instead of generating text; d1.2 adds audio input and five new task types: reading a value from the input, finding every span of a kind, locating boxes in images, estimating a number with a range, and multi-select choice. It bills input tokens only, with images at 1.5 tokens per 32x32-pixel patch. API-only and proprietary.