anthropic

Claude Haiku API Pricing (October 2026)

claude-haiku-4-5-20251001

Compare Claude Haiku API pricing, explore each model’s capabilities, and start building with one NanoGPT API key.

VisionNative PDF inputTool CallingStructured Output

Input / 1M tokens: $1.00
Output / 1M tokens: $5.00
Cache read / 1M tokens: $0.10
Cache creation / 1M tokens: $1.25
1-hour cache creation / 1M tokens: $2.00
Average output tokens: 248
Context: 200,000

Get API Key API documentation

Build with Claude Haiku 4.5

curl https://nano-gpt.com/api/v1/chat/completions \
  -H "Authorization: Bearer $NANOGPT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "claude-haiku-4-5-20251001",
  "messages": [
    {
      "role": "user",
      "content": "Hello!"
    }
  ]
}'

Set NANOGPT_API_KEY in your environment. Keep your API key on the server.

Haiku model pricing per 1M tokens

Find a modelProviderSort by

Input tokens per estimated query

USD per 1 million tokens · NanoGPT standard catalogue rates
Model / API ID	Input	Cache read	Cache creation	Output	Batch input	Batch output	Context	Average output tokens	Estimated request cost
Claude Haiku 4.5`claude-haiku-4-5-20251001`	$1.00	$0.10	$1.251-hour: $2.00	$5.00	$0.50	$2.50	200,000	248	$0.00224
Claude Haiku 4.5 Thinking`claude-haiku-4-5-20251001-thinking`	$1.00	$0.10	$1.251-hour: $2.00	$5.00	$0.50	$2.50	200,000	917	$0.00559
Claude Haiku 5.5`anthropic/claude-haiku-5.5`Higher rates above 100,000 input tokens	$0.10	$0.01	$0.1251-hour: $0.20	$0.50	$0.05	$0.25	1,000,000	1,066	$0.00063
Claude Haiku Latest`anthropic/claude-haiku-latest`Higher rates above 100,000 input tokens	$0.10	$0.01	$0.1251-hour: $0.20	$0.50	$0.05	$0.25	1,000,000	739	$0.00047

Base text-token rates. Batch columns show supported standard Batch API rates; long-context tiers and additional charges can apply. Long-context pricing, tool charges, service tiers, and provider selection can change the total. Cache creation is charged when writing tokens to the cache; cache read applies when reusing them. The default cache duration applies unless labeled otherwise. Unsupported cache rates are marked unavailable.

Estimated request cost combines your input-token count with the model’s average output, rounded to whole tokens, at current standard rates. It assumes uncached input and excludes tool charges. This is an estimate for that workload, not a measured average total bill.

Compare model intelligence and coding benchmarks

Compare models using published benchmark scores. Higher scores are better. Each chart uses one benchmark and source version; these are current comparisons, not scores recorded at launch. Ranking pairs each model’s standard and thinking evaluations on one row, Value plots score against the output token rate, and Generations tracks each family release by release. Thinking evaluations are labeled with their published settings, and evaluation effort can affect scores.

Benchmark

Artificial Analysis Intelligence Index

Source version not published · 23 models scored

Using cached Artificial Analysis scores. Recent models and evaluations may be missing.

Which model scores highest, and what does thinking add?

StandardThinkingWhat thinking adds

Hover, tap, or focus a marker for its evaluation settings and rate.

Source: Artificial Analysis. Missing scores mean no verified matching result is available here for this version; they do not mean the model was never evaluated. Token rates are NanoGPT standard catalogue rates.

View all benchmark scores and evaluation details

Artificial Analysis Intelligence Index (unversioned) by model · index points
Model	Score	Output / 1M
Claude Fable 5	49.6	$50
Claude Fable 5.1	53.4	$50
Claude Haiku 4.5	15.4	$5
Claude Haiku 4.5 Thinking	16.9	$5
Claude Haiku 5.5	43.4	$0.5
Claude 4 Opus	16.6	$75
Claude 4 Opus Thinking	Score unavailable	$75
Claude 4.1 Opus	18.6	$75
Claude 4.1 Opus Thinking	Score unavailable	$75
Claude 4.5 Opus	23.7	$25
Claude 4.5 Opus Thinking	29.1	$25
Claude 4.6 Opus	26.4	$25
Claude 4.6 Opus Thinking	31.9	$25
Claude 4.6 Opus Thinking Low	Score unavailable	$25
Claude 4.6 Opus Thinking Max	Score unavailable	$25
Claude 4.6 Opus Thinking Medium	Score unavailable	$25
Claude 4.7 Opus	30.9	$25
Claude 4.7 Opus Thinking	40.7	$25
Claude Opus 4.8	Score unavailable	$25
Claude Opus 4.8 Thinking	41.8	$25
Claude Opus 5	50.8	$25
Claude Opus 5.5	57.6	$20
Claude Sonnet 4.5	19.3	$15
Claude Sonnet 4.5 Thinking	20.7	$15
Claude Sonnet 4.6	24.7	$15
Claude Sonnet 4.6 Thinking	30.1	$15
Claude Sonnet 5	23.2	$10
Claude Sonnet 5 Thinking	38.2	$10
Claude Sonnet 5.5	Score unavailable	$10
Claude Sonnet 5.5 Thinking	56	$10

Artificial Analysis Coding Index (unversioned) by model · index points
Model	Score	Output / 1M
Claude Fable 5	76.5	$50
Claude Fable 5.1	81.6	$50
Claude Haiku 4.5	Score unavailable	$5
Claude Haiku 4.5 Thinking	43.9	$5
Claude Haiku 5.5	Score unavailable	$0.5
Claude 4 Opus	Score unavailable	$75
Claude 4 Opus Thinking	Score unavailable	$75
Claude 4.1 Opus	Score unavailable	$75
Claude 4.1 Opus Thinking	Score unavailable	$75
Claude 4.5 Opus	Score unavailable	$25
Claude 4.5 Opus Thinking	Score unavailable	$25
Claude 4.6 Opus	Score unavailable	$25
Claude 4.6 Opus Thinking	Score unavailable	$25
Claude 4.6 Opus Thinking Low	Score unavailable	$25
Claude 4.6 Opus Thinking Max	Score unavailable	$25
Claude 4.6 Opus Thinking Medium	Score unavailable	$25
Claude 4.7 Opus	Score unavailable	$25
Claude 4.7 Opus Thinking	73.6	$25
Claude Opus 4.8	Score unavailable	$25
Claude Opus 4.8 Thinking	74.3	$25
Claude Opus 5	78	$25
Claude Opus 5.5	Score unavailable	$20
Claude Sonnet 4.5	Score unavailable	$15
Claude Sonnet 4.5 Thinking	52.1	$15
Claude Sonnet 4.6	Score unavailable	$15
Claude Sonnet 4.6 Thinking	63	$15
Claude Sonnet 5	66.4	$10
Claude Sonnet 5 Thinking	71.5	$10
Claude Sonnet 5.5	Score unavailable	$10
Claude Sonnet 5.5 Thinking	Score unavailable	$10

anthropic

Claude Haiku 4.5

claude-haiku-4-5-20251001

Claude Haiku 4.5 matches Sonnet 4's coding quality while running about twice as fast at roughly one-third the price, making it ideal for low-latency chat assistants, support bots, and pair programming. Pair it with Sonnet 4.5 to plan complex work and fan out execution across parallel Haiku workers.

VisionNative PDF inputTool CallingStructured Output

Input / 1M tokens: $1.00
Output / 1M tokens: $5.00
Cache read / 1M tokens: $0.10
Cache creation / 1M tokens: $1.25
1-hour cache creation / 1M tokens: $2.00
Average output tokens: 248
Context: 200,000

Use this model API documentation

Build with Claude Haiku 4.5

curl https://nano-gpt.com/api/v1/chat/completions \
  -H "Authorization: Bearer $NANOGPT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "claude-haiku-4-5-20251001",
  "messages": [
    {
      "role": "user",
      "content": "Hello!"
    }
  ]
}'

Set NANOGPT_API_KEY in your environment. Keep your API key on the server.

anthropic

Claude Haiku 4.5 Thinking

claude-haiku-4-5-20251001-thinking

Claude Haiku 4.5 with extended thinking enabled for deeper planning, more deliberate coding, and multi-step reasoning without moving up to Sonnet pricing.

ReasoningVisionNative PDF inputTool CallingStructured Output

Input / 1M tokens: $1.00
Output / 1M tokens: $5.00
Cache read / 1M tokens: $0.10
Cache creation / 1M tokens: $1.25
1-hour cache creation / 1M tokens: $2.00
Average output tokens: 917
Context: 200,000

Use this model API documentation

Build with Claude Haiku 4.5 Thinking

curl https://nano-gpt.com/api/v1/chat/completions \
  -H "Authorization: Bearer $NANOGPT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "claude-haiku-4-5-20251001-thinking",
  "messages": [
    {
      "role": "user",
      "content": "Hello!"
    }
  ]
}'

Set NANOGPT_API_KEY in your environment. Keep your API key on the server.

anthropic

Claude Haiku Latest

anthropic/claude-haiku-latest

Compatibility alias that routes to the newest version of Claude Haiku. Currently routes to Claude Haiku 5.5 with adaptive thinking.

ReasoningVisionNative PDF inputTool CallingStructured Output

Input / 1M tokens: $0.10
Output / 1M tokens: $0.50
Cache read / 1M tokens: $0.01
Cache creation / 1M tokens: $0.125
1-hour cache creation / 1M tokens: $0.20
Average output tokens: 739
Context: 1,000,000

Use this model API documentation

Build with Claude Haiku Latest

curl https://nano-gpt.com/api/v1/chat/completions \
  -H "Authorization: Bearer $NANOGPT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "anthropic/claude-haiku-latest",
  "messages": [
    {
      "role": "user",
      "content": "Hello!"
    }
  ]
}'

Set NANOGPT_API_KEY in your environment. Keep your API key on the server.

anthropic

Claude Haiku 5.5

anthropic/claude-haiku-5.5

Claude Haiku 5.5 is Anthropic's fastest and most capable small model, designed for high-volume, cost-sensitive tasks such as summaries, classification, coding assistance, and live customer support. It delivers major gains over Haiku 4.5 while costing around 75% less to run.

ReasoningVisionNative PDF inputTool CallingStructured Output

Input / 1M tokens: $0.10
Output / 1M tokens: $0.50
Cache read / 1M tokens: $0.01
Cache creation / 1M tokens: $0.125
1-hour cache creation / 1M tokens: $0.20
Average output tokens: 1,066
Context: 1,000,000

Use this model API documentation

Build with Claude Haiku 5.5

curl https://nano-gpt.com/api/v1/chat/completions \
  -H "Authorization: Bearer $NANOGPT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "anthropic/claude-haiku-5.5",
  "messages": [
    {
      "role": "user",
      "content": "Hello!"
    }
  ]
}'

Set NANOGPT_API_KEY in your environment. Keep your API key on the server.

Estimate a request’s token cost

ModelInput tokensOutput tokens

$0.015

One request, standard uncached text tokens. Excludes tools and other charges. Known long-context rates apply automatically; invalid workloads or missing rates cannot be estimated.