How much does Claude Haiku 4.5 cost?

On the Respan gateway, Claude Haiku 4.5 costs from $0.50 per 1M input tokens and $2.50 per 1M output tokens on OpenRouter, after gateway discounts. Averaged across its 7 routes from 6 providers, it costs $0.93 per 1M input tokens and $4.64 per 1M output tokens.

Which providers serve Claude Haiku 4.5 on Respan?

Claude Haiku 4.5 is available through 7 routes from 6 providers: OpenRouter, Anthropic, DeepInfra, Inference.net, Perplexity, and Google Vertex AI. Google Vertex AI requires your own provider API key (BYOK).

What is Claude Haiku 4.5's context window?

Claude Haiku 4.5 accepts up to 200K tokens of context on its largest provider route.

How do I use Claude Haiku 4.5 with the Respan gateway?

Point any OpenAI-compatible SDK at https://api.respan.ai/api/ with your Respan API key and set the model to "anthropic/claude-haiku-4-5". To set it up with a coding agent, copy the agent setup prompt at the top of this page.

How fast is Claude Haiku 4.5?

Over the last 30 days of requests through the Respan gateway, the fastest provider was Anthropic at 78 tokens per second, and the lowest response latency was 1.3 seconds on Perplexity. Speed varies with request size and provider.

Where does Claude Haiku 4.5 rank on the Respan Index?

Claude Haiku 4.5 scores 17.6 out of 100 on the Respan Index v1.1, ranking #110 of 143 models. By category: Coding 6.7, Agents 16.8, Reasoning 12.7, Knowledge & language 13.2, and Human preference 48.3. The Respan Index combines independent and lab-reported benchmark results into one score, using benchmark data collected by llmmetric.com.

How does Claude Haiku 4.5 score on benchmarks?

Claude Haiku 4.5 has 17 published benchmark scores, including SWE-bench Verified 73.3%, BALROG 31.2%, and GPQA Diamond 71.2%. Each score links to where it was published; the data is collected by llmmetric.com.

Claude Haiku 4.5

Q: What is Claude Haiku 4.5?

Claude Haiku 4.5 is a model from Anthropic, released on Oct 15, 2025. Claude Haiku 4.5 is Anthropic’s fastest and most efficient model, delivering near-frontier intelligence at a fraction of the cost and latency of larger Claude models. $1 per million input tokens, $5 per million output tokens. 200,000 token context window, maximum output of 64,000 tokens. Higher uptime with 4 providers. Includes independent benchmarks from Artificial Analysis.

Claude Haiku 4.5 is Anthropic’s fastest and most efficient model, delivering near-frontier intelligence at a fraction of the cost and latency of larger Claude models. $1 per million input tokens, $5 per million output tokens. 200,000 token context window, maximum output of 64,000 tokens. Higher uptime with 4 providers. Includes independent benchmarks from Artificial Analysis.

Input / output, from: $0.50 / $2.50 per 1M
Weekly tokens / requests: 62.1M / 2.8K
Context: 200K tokens
Released: Oct 15, 2025

Try Respan for free

Agent setup

Paste into Claude Code, Cursor, or Codex

Help me set up the Respan gateway: fetch and follow https://www.respan.ai/docs/agent-setup.md. Use model "anthropic/claude-haiku-4-5".

from openai import OpenAI

client = OpenAI(

    base_url="https://api.respan.ai/api/",

    api_key="YOUR_RESPAN_API_KEY",

)

response = client.chat.completions.create(

    model="anthropic/claude-haiku-4-5",

    messages=[{"role": "user", "content": "Hello!"}],

)

print(response.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({

  baseURL: "https://api.respan.ai/api/",

  apiKey: "YOUR_RESPAN_API_KEY",

});

const response = await client.chat.completions.create({

  model: "anthropic/claude-haiku-4-5",

  messages: [{ role: "user", content: "Hello!" }],

});

console.log(response.choices[0].message.content);

curl https://api.respan.ai/api/chat/completions \

  -H "Authorization: Bearer YOUR_RESPAN_API_KEY" \

  -H "Content-Type: application/json" \

  -d '{

    "model": "anthropic/claude-haiku-4-5",

    "messages": [{"role": "user", "content": "Hello!"}]

}'

Read docs

Providers

7 routes from 6 providers, cheapest first, priced per 1M tokens after gateway discounts. BYOK routes use your own key.

Provider	Context	Input	Output	Cache read	Capabilities
OpenRouter`openrouter/anthropic/claude-haiku-4.5:batch`	200K	$0.50/M	$2.50/M	$0.05/M	+2
Anthropic`anthropic/claude-haiku-4-5`	200K	$1.00/M	$5.00/M	$0.10/M	+2
DeepInfra`deepinfra/anthropic/claude-haiku-4-5`	200K	$1.00/M	$5.00/M	$0.00/M	+2
Inference.net`inference_net/claude-haiku-4-5`	200K	$1.00/M	$5.00/M	$0.10/M
OpenRouter`openrouter/anthropic/claude-haiku-4.5`	200K	$1.00/M	$5.00/M	$0.10/M	+2
Perplexity`perplexity/anthropic/claude-haiku-4-5`	200K	$1.00/M	$5.00/M	$0.10/M
Google Vertex AIBYOK only`vertex_ai/claude-haiku-4-5@20251001`	200K	$1.00/M	$5.00/M	$0.10/M	+2

Is a provider down? OpenRouter status, Anthropic status, Perplexity status, Google AI status

Pricing

Averaged across 7 routes from 6 providers, in USD per 1M tokens after gateway discounts.

Average input: $0.93 per 1M
Average output: $4.64 per 1M
Average cache read: $0.08 per 1M
Average cache write: $0.80 per 1M

Cost by volume

Total cost from 0 to 1T tokens on the cheapest and most expensive provider.

Highest$1.00/MAnthropic, DeepInfra, Inference.net, OpenRouter, Perplexity, and Google Vertex AILowest$0.50/MOpenRouter

Throughput and latency

30-day averages from requests through the Respan gateway, 2 of 7 routes had traffic in that window. Under each figure, every model on the gateway by its average, this one in blue.

Output speed

Higher is better

78tokens/sbest, on Anthropic

Faster than 77% of the other 485 models on the gateway

Anthropic78 tok/s
Perplexity3.9 tok/s

Time to first token

Lower is better

0.7son Anthropic

Faster than 92% of the other 64 models on the gateway

Response latency

Lower is better

1.3sbest, on Perplexity

Faster than 35% of the other 501 models on the gateway

Perplexity1.3 s
Anthropic3.8 s

No traffic in the last 30 days on OpenRouter, DeepInfra, Inference.net, and Google Vertex AI.

Benchmarks

The Respan Index plus 17 published scores from 7 sources, each linked to where it was published.

17.6/ 100

Respan Index#110 of 143

143 ranked models. The shaded band is this model's 95% interval, 14.6 to 22.2. Qwen3-Max and DeepSeek R1 fall inside it, so the gap is within the margin of error.

110Claude Haiku 4.517.6
109Qwen3-Max17.9
111DeepSeek R117.4

Published scores

Benchmark	Score	Source	Published
SWE-bench VerifiedAnthropic	73.3%	Anthropic	Oct 15, 2025
WeirdMLHåvard Tveit Ihle	45.4%	Håvard Tveit Ihle	Oct 5, 2026
Terminal-Bench 4.0Anthropic	0%	Anthropic	Oct 7, 2026

Benchmark	Score	Source	Published
BALROGBALROG	31.2%	BALROG	Feb 24, 2026
OSWorld 2.1 (offline subset)Anthropic	15.7%	Anthropic	Oct 7, 2026
Vending-Bench 2Andon Labs	458.89 · #50 of 59	Andon Labs	Oct 5, 2026

Benchmark	Score	Source	Published
GPQA DiamondEpoch AI	71.2%	Epoch AI	Oct 5, 2026
Humanity's Last Exam (with tools)Anthropic	18.7%	Anthropic	Oct 7, 2026
Humanity's Last Exam (no tools)Anthropic	10.2%	Anthropic	Oct 7, 2026
Chess PuzzlesEpoch AI	8%	Epoch AI	Oct 5, 2026