Claude Haiku 3.5
Retired Anthropic model, API ID claude-3-5-haiku-20241022.
text → text · made by Anthropic
$0.8→$4per Mtok in / out
Anthropic · 3 sellers
Sold by 4 ways
| Seller | Lane | Rate |
|---|---|---|
| Anthropic api | batch | $0.4per Mtok in$2per Mtok outplatform.claude.com · read 2026-08-24 |
| AWS Bedrock cloud | standard | $0.8per Mtok in$4per Mtok out$0.08per Mtok cached$1per Mtok cache writeraw.githubusercontent.com · read 2026-09-16 |
| Anthropic api | standard | $0.8per Mtok in$4per Mtok out$0.08per Mtok cached$1per Mtok cache writeplatform.claude.com · read 2026-08-24 |
| DigitalOcean cloud | standard | $0.8per Mtok in$4per Mtok out$0.08per Mtok cached$1per Mtok cache writemodels.dev · read 2026-09-16 |
Measured 18 standings
| Place | Board | Metric | Score |
|---|---|---|---|
| 10thof 15 | Cad eval — Epoch AI | Overall pass (%) | 0.32 |
| 24thof 36 | Balrog — Epoch AI | Average progress | 0.193 |
| 35thof 49 | Lech mazur writing — Epoch AI | Mean score | 7.35 |
| 40thof 54 | Live bench — Epoch AI | Global average | 43.45 |
| 50thof 69 | Aider polyglot coding leaderboard | Percent correct | 28 |
| 54thof 72 | Aider polyglot — Epoch AI | Percent correct | 28 |
| 54thof 217 | Mmlu — Epoch AI | EM | 0.743 |
| 67thof 108 | Math level 5 — Epoch AI | mean_score | 46.356 |
| 97thof 101 | FrontierMath (Tiers 1-3) | mean_score | 0.345 |
| 114thof 152 | LMArena · Vision | Score (Elo) | 1127 |
| 125thof 161 | Dtbench — Epoch AI | Accuracy | 56.66 |
| 146thof 170 | Weirdml — Epoch AI | Accuracy | 30.73 |
| 147thof 177 | Critpt — Epoch AI | Accuracy | 0 |
| 154thof 168 | Scicode — Epoch AI | Score | 27.431 |
| 239thof 402 | LMArena · Text | Score (Elo) | 1324 |
| 263rdof 290 | OTIS Mock AIME 2024-2025 | mean_score | 4.306 |
| 270thof 313 | GPQA Diamond | mean_score | 38.131 |
| 545thof 656 | Epoch capabilities index — Epoch AI | ECI Score | 127.15 |
About
Claude Haiku 3.5 — a text model from Anthropic, sold by 3 companies from $0.8 in and $4 out per million tokens, placed 54th of 217 on Mmlu — Epoch AI.
It takes text and returns text, with a context window of 200,000 tokens. It was published in October 2024. Its sellers say it can call a tool. The catalogue files it under reasoning and code. Three companies sell it. The cheapest is $0.8 in and $4 out per million tokens at AWS Bedrock. Beside the standard rate there is a batch lane. It has been measured on 18 boards, and stands best at 54th of 217 on Mmlu — Epoch AI.
Every current figure
- Maker
- Anthropic
- Register
- model
- Takes
- text
- Returns
- text
- Context
- 200,000 tokens
- Longest answer
- 64,000 tokens
- Published
- October 2024
- Licence
- not read
- Sellers
- 3
- Price
- $0.8 in and $4 out per million tokens — AWS Bedrock
- Boards
- 18
- Best place
- 54th of 217 — Mmlu — Epoch AI
Known as 1 name
Claude 3.5 Haiku