Claude Haiku 4.5

Snowflake Cortex chat

Claude Haiku 4.5 is a Snowflake Cortex chat model.It supports a 200,000-token context windowwith up to 16,384 output tokens.Input is priced at $1.00/M tokens and output at $5.00/M tokens. Capabilities include function calling, vision, prompt caching. Route Claude Haiku 4.5 via Future AGI's Agent Command Center for unified observability, caching, and 15 routing strategies including cost-optimized fallback.

Pricing source: litellm Last verified: Aug 6, 2026 View source ↗
Cost calculator

Estimate Claude Haiku 4.5 spend

Pick a workload, fine-tune the sliders, and see the monthly bill.

~3K in / ~400 out · 5K req/day
3,000
0200,000
400
016,384
5,000
01,000,000
cached @ $0.1000/M
Per request
$0.005000
in $0.003000 · out $0.002000
Per day
$25.00
5,000 requests
Per month
$761
152,188 requests

Estimate uses $1.00/M input · $5.00/M output. Provider pricing changes. Production costs vary with retries, streaming overhead, and tool-call rounds.
Want this for free? Cache + route via Agent Command Center — first 100K requests and 100K cache hits free every month.

Pricing

Per-token rates, expressed in USD per 1M tokens. Verified Aug 6, 2026.

Input $1.00/M
Output $5.00/M
Blended (3:1 in:out) $2.00/M
Cross-model comparable price
Cached input $0.1000/M

Context & limits

Context window
200,000 tokens
Max input
200,000 tokens
Max output
16,384 tokens
Modalities
vision, text

Capabilities

  • Function calling ✓ supported
  • Parallel tool calls — not advertised
  • Vision input ✓ supported
  • Audio input — not advertised
  • Audio output — not advertised
  • PDF input — not advertised
  • Streaming ✓ supported
  • Structured output ✓ supported
  • Prompt caching ✓ supported
  • Reasoning — not advertised

Strengths & caveats — when to pick Claude Haiku 4.5

Data-driven from Claude Haiku 4.5's percentile within chat peers.

Where it's strong

  • +agentic workflows that depend on reliable tool calls
  • +document, chart, and screenshot understanding
  • +repeated long-context prompts (caching saves up to 90% on input cost)

Watch out for

  • No major caveats flagged from public spec.

Benchmark scores

Reported public benchmark numbers. Each row links to the source.

HumanEvalcode· 0-shot
Captured Aug 8, 2026
Chatbot Arena ELOgeneral· overall
Captured Aug 8, 2026
BFCL v3agent· multi-turn
Captured Aug 8, 2026
MMLU-Proreasoning· 0-shot
Captured Aug 8, 2026
GPQA Diamondreasoning· 0-shot CoT
Captured Aug 8, 2026
SWE-bench Verifiedagent· agentic
Captured Aug 8, 2026
Try it

Call Claude Haiku 4.5 via Agent Command Center

One OpenAI-compatible endpoint. Routing, fallback, semantic caching, guardrails, and cost tracking come along for the ride. First 100K requests + 100K cache hits free every month.

SDK
Native Future AGI client (agentcc / @agentcc/client). Per-call metadata — provider, cost, latency, cache hit, request id — is returned on x-agentcc-* response headers, so any HTTP client can read it.
# Claude Haiku 4.5 via the Agent Command Center Python SDK
# pip install agentcc
import os
from agentcc import AgentCC

client = AgentCC(
    api_key=os.environ["AGENTCC_API_KEY"],   # from app.futureagi.com → Settings → API Keys
    base_url="https://gateway.futureagi.com/v1",
)

resp = client.chat.completions.create(
    model="snowflake-cortex/claude-haiku-4-5",
    messages=[{"role": "user", "content": "Hello, Claude Haiku 4.5!"}],
)

print(resp.choices[0].message.content)
print(f"Tokens: {resp.usage.total_tokens}")

# Per-call gateway metadata is returned on x-agentcc-* response headers.
# When you need it programmatically, use .with_raw_response to get them:
raw = client.chat.completions.with_raw_response.create(
    model="snowflake-cortex/claude-haiku-4-5",
    messages=[{"role": "user", "content": "Same call, but I want the headers."}],
)
print("Provider:", raw.headers.get("x-agentcc-provider"))
print("Latency:", raw.headers.get("x-agentcc-latency-ms"), "ms")
print("Cost:   ", raw.headers.get("x-agentcc-cost"), "USD")
print("Cache:  ", raw.headers.get("x-agentcc-cache"))
Set AGENTCC_API_KEY with a key fromapp.futureagi.com.Gateway docs ↗
Advanced: fallback + cache config (YAML)
strategy: cost-optimized
targets:
  - model: claude-haiku-4-5
    provider: snowflake-cortex
    weight: 80
fallbacks:
  - model: claude-opus-4-5
    provider: azure-ai-foundry
  - model: claude-opus-4-5-20251101
    provider: anthropic
guardrails: [pii, prompt-injection, secrets]
cache: { exact: true, semantic: true }

Same model on other providers

claude-haiku-4-5 is also available via 4 other routes. Pricing, regions, and capabilities can differ — compare before routing production traffic.

ProviderInput / 1MOutput / 1MVerified
Azure AI Foundry$1.00/M$5.00/MAug 6, 2026
Anthropic$1.00/M$5.00/MAug 6, 2026
GitHub CopilotAug 6, 2026
Google Vertex AI$1.00/M$5.00/MAug 6, 2026

Compare with similar models

Grouped by Chatbot Arena tier (Claude Haiku 4.5 sits at 1412 ELO).

FAQ

How much does Claude Haiku 4.5 cost?

Input is priced at $1.00 per 1M tokens and output at $5.00 per 1M tokens (Snowflake Cortex, last verified Aug 6, 2026).

What is the context window of Claude Haiku 4.5?

Claude Haiku 4.5 supports a 200,000-token context window with up to 16,384 output tokens.

Does Claude Haiku 4.5 support function calling?

Yes — Claude Haiku 4.5 supports function (tool) calling.

Is Claude Haiku 4.5 good for production?

Claude Haiku 4.5 is well-suited for agentic workflows that depend on reliable tool calls and document, chart, and screenshot understanding.

How can I route to Claude Haiku 4.5 with fallback?

Use Agent Command Center: a single OpenAI-compatible endpoint that supports cost-optimized routing, latency-aware retries, model fallback, and shadow traffic. Configure once, swap models without app changes.

Useful links for Claude Haiku 4.5

Official sources, independent benchmarks, and pricing aggregators — no random search-engine guesses.