Claude 3 Haiku
Claude 3 Haiku is Anthropic's fastest, most compact model for near-instant responsiveness.
text + image → text · made by Anthropic
Sold by 6 ways
| Seller | Lane | Rate |
|---|---|---|
| AWS Bedrock cloud | in-region | $0.25per Mtok inpricing.us-east-1.amazonaws.com · read 2026-08-24 |
| AWS Bedrock cloud | standard | $0.3per Mtok in$1.5per Mtok out$0.03per Mtok cached$0.38per Mtok cache writeraw.githubusercontent.com · read 2026-09-16 |
| poe aggregator | standard | $0.21per Mtok in$1.1per Mtok out$0.021per Mtok cachedmodels.dev · read 2026-09-01 |
| Nous Research aggregator | standard | $0.25per Mtok in$1.25per Mtok out$0.03per Mtok cached$0.3per Mtok cache writeinference-api.nousresearch.com · read 2026-09-12 |
| OpenRouter aggregator | standard | $0.25per Mtok in$1.25per Mtok out$0.03per Mtok cached$0.3per Mtok cache writeopenrouter.ai · read 2026-08-24 |
| Vercel AI Gateway aggregator | standard | $0.25per Mtok in$1.25per Mtok out$0.03per Mtok cached$0.3per Mtok cache writeai-gateway.vercel.sh · read 2026-08-25 |
Measured 15 standings
| Place | Board | Metric | Score |
|---|---|---|---|
| 12thof 26 | Science qa — Epoch AI | Score | 72 |
| 15thof 15 | Cad eval — Epoch AI | Overall pass (%) | 0.12 |
| 28thof 28 | LiveCodeBench Leaderboard (default window 8/1/2024–5/1/2025, 454 problems) | Pass@1 | 20.2 |
| 55thof 137 | Wino grande — Epoch AI | Accuracy | 74.2 |
| 56thof 217 | Mmlu — Epoch AI | EM | 0.738 |
| 78thof 81 | Forecastbench — Epoch AI | Overall score | 53.2 |
| 94thof 108 | Math level 5 — Epoch AI | mean_score | 14.879 |
| 113thof 123 | Lmca — Epoch AI | Score | 8.84 |
| 141stof 152 | LMArena · Vision | Score (Elo) | 1000 |
| 145thof 161 | Dtbench — Epoch AI | Accuracy | 50.13 |
| 165thof 170 | Weirdml — Epoch AI | Accuracy | 9.84 |
| 275thof 313 | GPQA Diamond | mean_score | 36.301 |
| 277thof 290 | OTIS Mock AIME 2024-2025 | mean_score | 1.806 |
| 303rdof 402 | LMArena · Text | Score (Elo) | 1261 |
| 592ndof 656 | Epoch capabilities index — Epoch AI | ECI Score | 118.27 |
About
Claude 3 Haiku — a text model from Anthropic, sold by 5 companies from $0.21 in and $1.1 out per million tokens, placed 56th of 217 on Mmlu — Epoch AI.
It takes text and images and returns text, with a context window of 200,000 tokens. It was published in March 2024, trained on material up to August 2023. Its sellers say it can call a tool. The catalogue files it under extract, reasoning, code and agents. Five companies sell it. The cheapest is $0.21 in and $1.1 out per million tokens at poe. Beside the standard rate there is a regional lane. It has been measured on 15 boards, and stands best at 56th of 217 on Mmlu — Epoch AI.
Every current figure
- Maker
- Anthropic
- Register
- model
- Takes
- text + image
- Returns
- text
- Context
- 200,000 tokens
- Longest answer
- 4,096 tokens
- Published
- March 2024
- Knowledge to
- August 2023
- Licence
- not read
- Sellers
- 5
- Maker's own price
- not read
- Price
- $0.21 in and $1.1 out per million tokens — poe
- Boards
- 15
- Best place
- 56th of 217 — Mmlu — Epoch AI
Known as 3 names
Claude-3-HaikuClaude-Haiku-3anthropic/claude-3-haiku