Claude Haiku 4.5 vs Claude Haiku 5.5
Claude Haiku 5.5 is cheaper, larger and much stronger than Claude Haiku 4.5. It costs $0.10 input and $0.50 output per 1M tokens against $1 and $5, reads 1M tokens instead of 200K, and scored 39.2% on Terminal-Bench 4.0, where the published score for Haiku 4.5 is 0%.
A year between them
Anthropic released Haiku 4.5 on 15 October 2025 and Haiku 5.5 on 7 October 2026. In that year the context window grew five times, from 200K to 1M tokens, and the output limit doubled, from 64K to 128K. Input and output prices fell to a tenth, and cached input from $0.10 to $0.01.
The long-prompt rate
Haiku 5.5 has a second price. A request with more than 100K prompt tokens costs $0.50 input and $2.50 output per 1M. Haiku 4.5 has one price, but it cannot take a prompt over 200K at all.
So the saving depends on prompt size. Our feature session, with 80K-token prompts, costs $5.40 on Haiku 4.5 and $0.54 on Haiku 5.5. A long agent day at 150K tokens per prompt costs $48.75 against $24.38: half, not a tenth.
Scores
On Terminal-Bench 4.0 Haiku 4.5 is listed at 0% and Haiku 5.5 at 39.2%; the 0% is the figure in the table we cite, not a missing value. On Humanity's Last Exam with tools it is 18.7% against 57.4%. Haiku 4.5 has no published FrontierCode 1.1 score; Haiku 5.5 scored 46.4%.
Side by side
| Claude Haiku 4.5 | Claude Haiku 5.5 | |
|---|---|---|
| Maker | Anthropic | Anthropic |
| Released | 15 October 2025[1] | 7 October 2026[2] |
| Context window | 200K tokens[1] | 1M tokens[2] |
| Max output | 64K tokens[1] | 128K tokens[2] |
| Input price | $1 per 1M tokens[1] | $0.10 per 1M tokens[2] |
| Output price | $5 per 1M tokens[1] | $0.50 per 1M tokens[2] |
| Cached input | $0.10 per 1M tokens[3] | $0.01 per 1M tokens[3] |
| Long prompts | One price | Over 100K: $0.50 in, $2.50 out[2] Prompts over 100,000 tokens are billed at the higher rate. |
| Price change | None announced | None announced |
| Open weights | No[1] Offered through the Claude API and cloud platforms; no weights are published. | No[2] Offered through the Claude API and cloud platforms; no weights are published. |
| Reads | text, image[1] | text, image[2] |
Benchmarks
Scores as the makers published them, each with its source and date. Makers run these at different settings, so read a gap of a few points as a tie.
| Benchmark | Claude Haiku 4.5 | Claude Haiku 5.5 |
|---|---|---|
| Terminal-Bench 4.0Agentic coding tasks in a real terminal. | 0%[4] | 39.2%[4] |
| FrontierCode 1.1 (Main)Hard software engineering tasks. | Not published | 46.4%[4] |
| Humanity's Last Exam (with tools)Expert-level questions across many fields, tools allowed. | 18.7%[4] | 57.4%[4] |
What a coding session costs
Our own estimate from the list prices above, for three sessions we define: a coding agent resends the conversation with every call, so prompts are large and grow. "Cached" assumes 90% of prompt tokens are read from the prompt cache and leaves out cache write fees. A long-prompt price applies when a call's prompt is over its threshold. Work out your own mix with the token cost calculator.
| Session | Claude Haiku 4.5 | Claude Haiku 4.5 cached | Claude Haiku 5.5 | Claude Haiku 5.5 cached |
|---|---|---|---|---|
| Quick fix10 calls of 25K prompt and 1.5K output tokens | $0.33 | $0.12 | $0.03 | $0.01 |
| Feature session60 calls of 80K prompt and 2K output tokens | $5.40 | $1.51 | $0.54 | $0.15 |
| Long agent day300 calls of 150K prompt and 2.5K output tokens | $48.75 | $12.30 | $24.38 | $6.15 |
Value for money and release gap
Claude Haiku 5.5 came out 1 year later (357 days). Price per point divides our feature-session cost by a published score: lower means more score for the money. It is a rough lens, not a ranking.
| Measure | Claude Haiku 4.5 | Claude Haiku 5.5 |
|---|---|---|
| Input tokens per $1 | 1M | 10M |
| Filling the whole context window once | $0.20 | $0.50 |
| Feature session per point of Terminal-Bench 4.0 | n/a | $0.01 |
| Feature session per point of Humanity's Last Exam (with tools) | $0.29 | $0.0094 |
Questions
Is Haiku 5.5 cheaper than Haiku 4.5?
Yes. For prompts up to 100K tokens it costs a tenth: $0.10 against $1 per 1M input tokens and $0.50 against $5 per 1M output. Above 100K, Haiku 5.5 charges $0.50 and $2.50.
Did Haiku 4.5 really score 0% on Terminal-Bench 4.0?
That is the figure in the source we cite, which is linked from the benchmark table. Haiku 5.5 scored 39.2% on the same benchmark.
How much bigger is the Haiku 5.5 context window?
Five times: 1M tokens against the 200K of Haiku 4.5. The output limit doubled as well, from 64K to 128K tokens.
Most compared
Any other two models: pick them in the lists at the top of the page.
- Claude Haiku 5.5 vs Claude Sonnet 5.5
- Claude Haiku 5.5 vs GPT-6 Luna
- Claude Haiku 5.5 vs Gemini 3.8 Flash
- Claude Opus 5.5 vs Claude Sonnet 5.5
- Claude Fable 5.1 vs Claude Opus 5.5
- Claude Opus 5 vs Claude Opus 5.5
- Claude Sonnet 5 vs Claude Sonnet 5.5
- Claude Fable 5 vs Claude Fable 5.1
- Claude Opus 5.5 vs GPT-6.1 Sol
- Claude Sonnet 5.5 vs GPT-6.1 Sol
- Claude Opus 5.5 vs GPT-5.6 Sol
- Claude Fable 5.1 vs GPT-6 Astra
- Claude Opus 5.5 vs GPT-6 Astra
- GPT-6.1 Sol vs GPT-6 Sol
- GPT-6.1 Sol vs GPT-6 Astra
- Claude Opus 5.5 vs Gemini 4 Argon
- Claude Sonnet 5.5 vs Gemini 4 Argon
- Claude Opus 5.5 vs Gemini 3.1 Pro Preview
- Gemini 4 Argon vs GPT-6.1 Sol
- Gemini 3.1 Pro Preview vs Gemini 4 Argon
- Claude Opus 5.5 vs Grok 4.7
- Claude Opus 5.5 vs DeepSeek V4 Pro
- Claude Sonnet 5.5 vs Kimi K3
- Claude Sonnet 5.5 vs GLM-5.3
More tools
Sources
Numbers from the makers' own pages, checked by hand, last read . List prices in US dollars before tax. Spotted a mistake? office@shuriken.sh.
Made by Shuriken, the harness for coding agents. Its hot swapping moves your chats to another of your accounts or API keys when one nears its limit. Use only accounts you own and are entitled to use.
Try Shuriken on your own computerYou're signed in
Add your Claude, Codex and Gemini accounts once. Shuriken switches between them so your agents keep working.Your accounts, projects and chats are in the app.
Coming soon