Which is cheaper, ERNIE 4.5 VL 424B A47B or Claude 3 Haiku?

Claude 3 Haiku is the cheaper of the two: one 1,000-word request costs about 566 toman.

Which one has the larger context window?

Claude 3 Haiku, with 200,000 tokens.

ERNIE 4.5 VL 424B A47B vs Claude 3 Haiku

Both sit behind the same key and the same code on uttapen — the only thing that changes is the model string, so you can run each of them against your own workload without touching anything else.

Feature	ERNIE 4.5 VL 424B A47B	Claude 3 Haiku
Provider	Baidu	Anthropic
Model id	baidu/ernie-4.5-vl-424b-a47b	anthropic/claude-3-haiku
Context	123,000 tokens	200,000 tokens
Max output	16,000 tokens	4,096 tokens
Input / 1M tokens	121,853 toman	72,531 toman
Output / 1M tokens	362,656 toman	362,656 toman
≈ one 1,000-word request	630 toman	566 toman
Input caching	No	8,704 toman / 1M
Image input	Yes	Yes
File input	No	No
Tool calling	No	Yes
JSON output	No	No
Reasoning mode	Yes	No

Which one for what?

ERNIE 4.5 VL 424B A47B

Multi-step problems, maths, and tracking down a bug
Reading screenshots, invoices, and scanned forms

630 toman per 1,000 words · Model page

Claude 3 Haiku

Whole documents or a codebase in a single request
Reading screenshots, invoices, and scanned forms
Agents that call your own APIs and database

566 toman per 1,000 words · Model page

Run both with the same code

Swap the model value between the two ids and send the same request twice; what each answer cost comes back in the X-Uttapen-Cost-Toman header.

curl https://api.uttapen.ir/v1/chat/completions \
  -H "Authorization: Bearer sk-up-…" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "baidu/ernie-4.5-vl-424b-a47b",
    "messages": [{"role": "user", "content": "Introduce yourself in one sentence."}],
    "stream": true
  }'

from openai import OpenAI

client = OpenAI(
    base_url="https://api.uttapen.ir/v1",
    api_key="sk-up-…",
)

stream = client.chat.completions.create(
    model="baidu/ernie-4.5-vl-424b-a47b",
    messages=[{"role": "user", "content": "Introduce yourself in one sentence."}],
    stream=True,
)
for chunk in stream:
    print(chunk.choices[0].delta.content or "", end="", flush=True)

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.uttapen.ir/v1",
  apiKey: "sk-up-…",
});

const stream = await client.chat.completions.create({
  model: "baidu/ernie-4.5-vl-424b-a47b",
  messages: [{ role: "user", content: "Introduce yourself in one sentence." }],
  stream: true,
});
for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content ?? "");
}

<?php
// composer require openai-php/client guzzlehttp/guzzle
$client = OpenAI::factory()
    ->withBaseUri('https://api.uttapen.ir/v1')
    ->withApiKey('sk-up-…')
    ->make();

$result = $client->chat()->create([
    'model' => 'baidu/ernie-4.5-vl-424b-a47b',
    'messages' => [['role' => 'user', 'content' => 'Introduce yourself in one sentence.']],
]);
echo $result->choices[0]->message->content;

package main

import (
	"context"
	"fmt"

	"github.com/openai/openai-go"
	"github.com/openai/openai-go/option"
)

func main() {
	client := openai.NewClient(
		option.WithBaseURL("https://api.uttapen.ir/v1"),
		option.WithAPIKey("sk-up-…"),
	)
	resp, err := client.Chat.Completions.New(context.Background(), openai.ChatCompletionNewParams{
		Model:    "baidu/ernie-4.5-vl-424b-a47b",
		Messages: []openai.ChatCompletionMessageParamUnion{openai.UserMessage("Introduce yourself in one sentence.")},
	})
	if err != nil {
		panic(err)
	}
	fmt.Println(resp.Choices[0].Message.Content)
}

model = "anthropic/claude-3-haiku"

More comparisons: all pairs · model rankings · full catalogue

base_url = https://api.uttapen.ir/v1