Back to guides
Pricing·June 8, 2026·Updated July 2, 2026·9 min read

Anthropic Claude API pricing 2026 — Fable 5, Sonnet 5, Opus & Haiku costs, examples, cheaper access

Claude is the default for reasoning, coding and agents. Here are the current Claude API prices per model — from Haiku to the frontier Fable 5, 30–60% below Anthropic's list — with worked examples, the official Anthropic reference, and the levers that cut a Claude bill the most.

This is Anthropic Claude API pricing for 2026: the per-model token rates, worked cost examples, and the levers that cut a Claude bill the most. Claude is the default choice for reasoning, coding and agentic work, and Kunavo undercuts Anthropic's list on every model — about 60% off the Claude 4.x generation and 30% off the new Claude 5 family (Fable 5, Sonnet 5) — behind one OpenAI-compatible API.

Rates last verified July 2, 2026. Kunavo's per-token prices below are read live from the model catalog, and the “Anthropic list” column tracks Anthropic's published rate — the official source is anthropic.com/pricing (developer detail at docs.anthropic.com).

Claude API pricing at a glance

Rates are per 1M tokens, in USD, as billed on Kunavo. The “Anthropic list” column is Anthropic's published rate for the same model.

ModelInput / 1MOutput / 1MAnthropic list (in / out)You save
claude-haiku-4-5$0.40$2.00$1.00 / $5.00~60%
claude-sonnet-4-6$1.20$6.00$3.00 / $15.00~60%
claude-sonnet-5$2.10$10.50$3.00 / $15.00~30%
claude-opus-4-7$2.00$10.00$5.00 / $25.00~60%
claude-fable-5$7.00$35.00$10.00 / $50.00~30%

Live rates always show on the pricing page and each model page. Haiku is the cheap workhorse, Sonnet the balanced default, Opus the heavy reasoner — and the new Claude 5 family adds Sonnet 5 (near-Opus coding at Sonnet cost) and Fable 5, Anthropic's most capable model for frontier reasoning and long-horizon agents.

Pricing note on Claude Sonnet 5: the “Anthropic list” column anchors to Anthropic's standard sticker price of $3.00 / $15.00 per 1M. Anthropic is running an introductory rate of $2 / $10 through August 31, 2026 — during that window the intro rate is below Kunavo's $2.10 / $10.50; after it ends, Kunavo stays ~30% under list. Claude Fable 5 has no intro pricing — $7.00 / $35.00 vs Anthropic's $10.00 / $50.00 is a straight ~30% saving from day one.

How Claude pricing works

You pay for input tokens (system prompt, context, messages) and output tokens (the completion). Output is billed at roughly 5× the input rate across the family, so the largest lever on cost is how much the model writes. The second lever is prompt caching: stable, repeated context can be cached and billed at a fraction of the normal input rate.

Worked cost examples

Real numbers at Kunavo's rates:

WorkloadTokens (in / out)ModelCost
Support ticket800 / 200Haiku 4.5$0.00072
RAG answer6,000 / 500Sonnet 4.6$0.0102
Coding assistant call10,000 / 1,500Sonnet 4.6$0.021
Coding agent turn10,000 / 1,500Sonnet 5$0.037
Long-context analysis50,000 / 2,000Opus 4.7$0.12
Frontier agent run50,000 / 2,000Fable 5$0.42

So 10,000 support tickets a day on Haiku is about $7.20/day. The math, runnable:

claude_cost.py
# Kunavo Claude rates (USD per 1M tokens): (input, output)
RATES = {
    "claude-haiku-4-5":  (0.40, 2.00),
    "claude-sonnet-4-6": (1.20, 6.00),
    "claude-sonnet-5":   (2.10, 10.50),
    "claude-opus-4-7":   (2.00, 10.00),
    "claude-fable-5":    (7.00, 35.00),
}

def cost(model: str, in_tokens: int, out_tokens: int) -> float:
    i, o = RATES[model]
    return in_tokens / 1_000_000 * i + out_tokens / 1_000_000 * o

print(cost("claude-haiku-4-5", 800, 200))      # support ticket -> $0.00072
print(cost("claude-sonnet-4-6", 6_000, 500))   # RAG answer     -> $0.0102
print(cost("claude-sonnet-5", 10_000, 1_500))  # coding agent   -> $0.037
print(cost("claude-opus-4-7", 50_000, 2_000))  # long analysis  -> $0.12
print(cost("claude-fable-5", 50_000, 2_000))   # frontier run   -> $0.42

Prompt caching — the biggest single saving

If your system prompt or retrieved context is large and stable, prompt caching bills cached input at 10% of the input rate. For a long, reused system prompt that is a 60–90% cut on the input line. It is one extra field on the request — see the Anthropic prompt caching deep dive for the exact pattern and the Messages API reference.

Kunavo pricing and Stripe billing

No subscription, no Anthropic Console billing setup. Top up a balance (Stripe or local payment methods) and calls draw down at the rates above. Pay-as-you-go from a $5 minimum top-up, the balance never expires, and larger top-ups carry bonus credit. The same balance covers Claude, Gemini, GPT, image, video and audio models — one wallet, one invoice.

Claude vs Gemini: the cost decision

Match the model to the job rather than defaulting to the most capable one:

NeedPickKunavo in / out per 1M
Cheapest viablegemini-2-5-flash$0.09 / $0.75
Cheap + Anthropic qualityclaude-haiku-4-5$0.40 / $2.00
Balanced reasoningclaude-sonnet-4-6$1.20 / $6.00
Near-Opus coding at Sonnet costclaude-sonnet-5$2.10 / $10.50
Hardest reasoning on a budgetclaude-opus-4-7$2.00 / $10.00
Frontier reasoning & agentsclaude-fable-5$7.00 / $35.00

See the Gemini and GPT pricing guides for the other providers, and the cost optimization guide for the difficulty-routing pattern in code.

Claude Code API pricing

Claude Code — Anthropic's agentic coding CLI — runs on these same Claude models, so its cost is ordinary Claude API token cost: the files and tool output it reads count as input tokens, and the edits and explanations it writes count as output tokens. Sonnet 4.6 ($1.20 / $6.00 per 1M) is the value driver, Sonnet 5 ($2.10 / $10.50) the newest near-Opus coder, and Opus 4.7 ($2.00 / $10.00) or the frontier Fable 5 ($7.00 / $35.00) handle the hardest refactors. Agentic sessions re-read context each turn, so they are token-heavy — which makes prompt caching the biggest lever, billing cached context at 10% of the input rate. Point Claude Code (or another agentic coding tool) at Kunavo's Anthropic-compatible Messages endpoint to pay the rates above — 30–60% under Anthropic's list depending on the model — instead of list price.

FAQ

Is the Claude API free?

Anthropic has no free Claude API tier — you pay per token from the first call. Kunavo is pay-as-you-go from a $5 minimum top-up, with per-token rates 30–60% under Anthropic's list depending on the model, and a balance that never expires.

How much does Claude Sonnet cost?

Claude Sonnet 5 is $2.10 per 1M input and $10.50 per 1M output on Kunavo — about 30% under Anthropic's $3.00 / $15.00 sticker. The previous-gen Sonnet 4.6 is $1.20 / $6.00, about 60% under list; a 6K-in / 500-out RAG answer on it costs about $0.01.

How much does the Claude Fable 5 API cost?

Claude Fable 5 — Anthropic's most capable model — lists at $10.00 input / $50.00 output per 1M tokens. On Kunavo it is $7.00 / $35.00, about 30% under list. A 50K-in / 2K-out frontier agent run costs about $0.48.

How do I reduce Claude API cost?

Prompt caching first (cached input at 10% of rate), then tier down to Haiku for simple tasks and cap output. Details in the cost optimization guide.

How much does the Claude Code API cost?

Claude Code runs on the standard Claude models, so it's billed as ordinary Claude token cost — Sonnet 4.6 at $1.20 / $6.00, Sonnet 5 at $2.10 / $10.50 and Opus 4.7 at $2.00 / $10.00 per 1M on Kunavo, 30–60% under Anthropic depending on the model. Prompt caching cuts the token-heavy agentic sessions most.

Does Kunavo support the Anthropic Messages API?

Yes — Claude is available through both the native Messages API (/v1/messages) and the OpenAI-compatible /v1/chat/completions endpoint. To start, see how to get a Claude API key.