Anthropic model product
Claude Haiku 3
Claude 3 Haiku is the fastest and most compact model in the Claude 3 family, designed for near-instant responsiveness.
Updated Aug 17, 2026. Default version: Claude 3 Haiku
LLMBoard Score0.0Claude 3 Haiku
Coverage80%10 benchmark families
Context window200KTokens
Official input priceN/AOfficial price unavailable
Claude Haiku 3 Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Claude Haiku 3 Benchmark Results
Benchmark scores for Claude 3 Haiku.
Sort by
Benchmark | Score | Rank | Participants | Percentile | Evidence | Evaluated |
|---|
| BenchmarkBIG-Bench Hard | Score73.7% | Rank10 | Participants21 | Percentile55.0% | EvidenceC | Evaluated |
| BenchmarkHellaSwag | Score85.9% | Rank11 | Participants27 | Percentile61.5% | EvidenceC | Evaluated |
| BenchmarkARC-C | Score89.2% | Rank12 | Participants34 | Percentile66.7% | EvidenceC | Evaluated |
| BenchmarkDROP | Score78.4% | Rank18 | Participants30 | Percentile41.4% | EvidenceC | Evaluated |
| BenchmarkMGSM | Score75.1% | Rank20 | Participants31 | Percentile36.7% | EvidenceC | Evaluated |
| BenchmarkGSM8k | Score88.9% | Rank29 | Participants48 | Percentile40.4% | EvidenceC | Evaluated |
| BenchmarkHumanEval | Score75.9% | Rank48 | Participants66 | Percentile27.7% | EvidenceC | Evaluated |
| BenchmarkMATH | Score38.9% | Rank67 | Participants71 | Percentile5.7% | EvidenceC | Evaluated |
| BenchmarkMMLU | Score75.2% | Rank75 | Participants101 | Percentile26.0% | EvidenceC | Evaluated |
| BenchmarkGPQA | Score33.3% | Rank222 | Participants239 | Percentile7.1% | EvidenceC | Evaluated |
Claude Haiku 3 Arena Results
Preference and agent-evaluation results for the default version.
Sort by
Arena | Category | Rank | Rating / score | Votes | Observations | Result date |
|---|
| Arenavision style control | Categoryoverall | Rank134 | Rating / score1001.0 | Votes13,380 | ObservationsN/A | Result date |
| Arenavision | Categoryoverall | Rank137 | Rating / score950.3 | Votes13,380 | ObservationsN/A | Result date |
| Arenatext style control | Categoryoverall | Rank290 | Rating / score1261.2 | Votes117,701 | ObservationsN/A | Result date |
| Arenatext | Categoryoverall | Rank294 | Rating / score1194.8 | Votes117,701 | ObservationsN/A | Result date |
Claude Haiku 3 Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- N/A
- Official provider
- N/A
- Lowest third-party
- From $0.25 input, $1.25 output per 1M via OpenRouter
- Tracked offerings
- 2
Sort by
Provider | Provider model ID | Region | Input / 1M | Output / 1M | Context | Updated |
|---|
| ProviderOpenRouter | Provider model IDanthropic/claude-3-haiku | Regionglobal | Input / 1M$0.25 | Output / 1M$1.25 | Context200K | Updated |
| ProviderHelicone | Provider model IDclaude-3-haiku-20240307 | Regionglobal | Input / 1M$0.25 | Output / 1M$1.25 | Context200K | Updated |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
Claude Haiku 3 Runtime Performance
Provider-specific output speed and catalog latency for Claude 3 Haiku. Runtime does not affect the capability score.
Sort by
Provider | Output Speed | Catalog Latency | Max Input | Max Output | Updated |
|---|
| ProviderBedrock | Output Speed104 tok/s | Catalog Latency0.5 s | Max Input200K | Max Output200K | Updated |
| ProviderAnthropic | Output Speed100 tok/s | Catalog Latency0.5 s | Max Input200K | Max Output200K | Updated |
| ProviderGoogle | Output Speed42 tok/s | Catalog Latency0.4 s | Max Input200K | Max Output200K | Updated |
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
Claude Haiku 3 Specifications
Technical details for the model's default version.
- Version
- Claude 3 Haiku
- Released
- Mar 13, 2024
- Knowledge cutoff
- Aug 31, 2023
- Parameters
- N/A
- Context window
- 200K
- Max output
- 4.1K
- Inputs
- image, pdf, text
- Outputs
- text
- Open weights
- No
- License
- Proprietary
Claude Haiku 3 Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
Sort by
Version | Released | LLMBoard | Parameters | Context | Max output | Open weights | License |
|---|
| VersionClaude 3 Haiku | Released | LLMBoard0.0 | ParametersN/A | Context200K | Max output4.1K | Open weightsNo | LicenseProprietary |
Claude Haiku 3 vs nearby models
Open a comparison with the three ranked models immediately above and below this model.
Models similar to Claude Haiku 3
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
AN
Claude Sonnet 3
Anthropic
AN
Claude Haiku 3.5
Anthropic
AN
Claude Opus 3
Anthropic
AN
Claude Sonnet 3.5
Anthropic
AN
Claude Sonnet 3.7
Anthropic
AN
Claude Sonnet 4
Anthropic
What is Claude Haiku 3?
Key information about Claude Haiku 3 and its available data.
Claude 3 Haiku is the fastest and most compact model in the Claude 3 family, designed for near-instant responsiveness. It excels at answering simple queries and requests with unmatched speed, making it ideal for seamless AI experiences that mimic human interactions.
Data as of 2026-08-17.
FAQ
Common questions about Claude Haiku 3.
When was Claude Haiku 3 released?
Claude Haiku 3's default version was released on Mar 13, 2024.
How much does Claude Haiku 3 cost?
No official standard PAYG price is currently available for Claude Haiku 3. The lowest tracked third-party offer starts at $0.25 input and $1.25 output via OpenRouter.
Who created Claude Haiku 3?
Claude Haiku 3 was created by Anthropic.
What is the context window for Claude Haiku 3?
The default version has a 200K token context window.
Is Claude Haiku 3 open weight?
No. The default version is not marked as having publicly available weights.
How many API providers offer Claude Haiku 3?
2 provider offerings are linked to the default version.
What models should I compare Claude Haiku 3 with?
Nearby ranked alternatives include Gemma 3n E2B LiteRT, MedGemma 4B IT, Ministral 8B.