Skip to content
llmwise
Sign inStart free

Claude Haiku 4.5 · Data analysis

Claude Haiku 4.5 for data analysis: our test runs

Claude Haiku 4.5 ran our 5 data analysis prompts in llmwise, through the same pipeline your messages take. Here's every reply as it came, whether it passed, what it cost and what it counts as on Pro, next to the other models.

Start freeSee pricing

Based on 5 of our test runs on , through OpenRouter with the app's own prompt and settings. Updated .

Short answer

In our test runs on September 27, 2026, Claude Haiku 4.5 passed 3 of 5 of our data analysis prompts (1 of the 2 hard ones): 20th of 20 models. Each reply counted as 1 message on Pro, where it has 250 a month, and cost $0.0025 to run on average.

Prompt by prompt

Each of the 5 data analysis prompts, and how Claude Haiku 4.5's reply did.

Claude Haiku 4.5 on each data analysis prompt
PromptResultCostTime
The region with the most revenueEverydayPassed: Final answer North region brought in the most revenue with 1,085.: right.$0.00262.7 s
Average order value in AugustEverydayFailed: Final answer 275.00; expected 300.$0.00232.1 s
Revenue change from July to AugustEverydayPassed: Final answer 67.4%: right.$0.00224.2 s
A median, filtered two waysHardPassed: Final answer 6.25: right.$0.00162.0 s
Correlation between ad spend and sign-upsHardFailed: Final answer 0.99; expected 0.97.$0.00375.0 s

Every data analysis reply from Claude Haiku 4.5

  1. The region with the most revenue

    An everyday prompt: “The region with the most revenue”, in full.

    Passed: Final answer North region brought in the most revenue with 1,085.: right.

    I'll calculate the revenue for each order and sum by region.

**Revenue by order (quantity × price):**

| Order | Region | Revenue |
|-------|--------|---------|
| 1001 | West | 2 × $45 = $90 |
| 1002 | East | 1 × $120 = $120 |
| 1003 | West | 1 × $310 = $310 |
| 1004 | North | 4 × $45 = $180 |
| 1005 | East | 3 × $45 = $135 |
| 1006 | North | 2 × $120 = $240 |
| 1007 | West | 2 × $120 = $240 |
| 1008 | East | 1 × $310 = $310 |
| 1009 | North | 2 × $310 = $620 |
| 1010 | West | 5 × $45 = $225 |
| 1011 | East | 3 × $120 = $360 |
| 1012 | North | 1 × $45 = $45 |

**Total revenue by region:**
- **West:** $90 + $310 + $240 + $225 = $865
- **East:** $120 + $135 + $310 + $360 = $925
- **North:** $180 + $240 + $620 + $45 = $1,085

Final answer: North region brought in the most revenue with $1,085.

    675 tokens in, 390 out · 2.7 s · $0.0026 · 1 message on Pro · answered by anthropic/claude-haiku-4.5 via Amazon Bedrock ·

  2. Average order value in August

    An everyday prompt: “Average order value in August”, in full.

    Failed: Final answer 275.00; expected 300.

    I'll find the average revenue per order for August 2026.

**August 2026 orders:**

| order_id | date | quantity | price | revenue |
|----------|------|----------|-------|---------|
| 1007 | 2026-08-02 | 2 | 45.00 | 90.00 |
| 1008 | 2026-08-06 | 1 | 310.00 | 310.00 |
| 1009 | 2026-08-11 | 2 | 310.00 | 620.00 |
| 1010 | 2026-08-15 | 5 | 45.00 | 225.00 |
| 1011 | 2026-08-19 | 3 | 120.00 | 360.00 |
| 1012 | 2026-08-27 | 1 | 45.00 | 45.00 |

**Calculation:**
- Total revenue: 90 + 310 + 620 + 225 + 360 + 45 = 1,650.00
- Number of orders: 6
- Average revenue per order: 1,650.00 ÷ 6 = 275.00

Final answer: 275.00

    683 tokens in, 326 out · 2.1 s · $0.0023 · 1 message on Pro · answered by anthropic/claude-haiku-4.5 via Amazon Bedrock ·

  3. Revenue change from July to August

    An everyday prompt: “Revenue change from July to August”, in full.

    Passed: Final answer 67.4%: right.

    I'll calculate the revenue for each order, then sum by month.

**July 2026 orders:**
- 1001: 2 × $45 = $90
- 1002: 1 × $120 = $120
- 1003: 1 × $310 = $310
- 1004: 4 × $45 = $180
- 1005: 3 × $45 = $135
- 1006: 2 × $120 = $240

**July total: $1,075**

**August 2026 orders:**
- 1007: 2 × $120 = $240
- 1008: 1 × $310 = $310
- 1009: 2 × $310 = $620
- 1010: 5 × $45 = $225
- 1011: 3 × $120 = $360
- 1012: 1 × $45 = $45

**August total: $1,800**

**Percentage change:**
$$\frac{1,800 - 1,075}{1,075} × 100 = \frac{725}{1,075} × 100 = 67.4\%$$

Final answer: 67.4%

    688 tokens in, 309 out · 4.2 s · $0.0022 · 1 message on Pro · answered by anthropic/claude-haiku-4.5 via Amazon Bedrock ·

  4. A median, filtered two ways

    A hard prompt: “A median, filtered two ways”, in full.

    Passed: Final answer 6.25: right.

    I'll filter for high-priority Billing team tickets and find the median.

**Filtering the data:**

High-priority tickets handled by Billing team:
- T1: 5.5 hours
- T4: 2 hours
- T6: 9 hours
- T9: 7 hours
- T10: 26 hours
- T12: 4 hours

**Sorting by hours_to_resolve:**
2, 4, 5.5, 7, 9, 26

**Calculating median:**
With 6 values (even number), the median is the average of the 3rd and 4th values:
- 3rd value: 5.5
- 4th value: 7
- Median: (5.5 + 7) / 2 = 12.5 / 2 = 6.25

Final answer: 6.25

    550 tokens in, 216 out · 2.0 s · $0.0016 · 1 message on Pro · answered by anthropic/claude-haiku-4.5 via Amazon Bedrock ·

  5. Correlation between ad spend and sign-ups

    A hard prompt: “Correlation between ad spend and sign-ups”, in full.

    Failed: Final answer 0.99; expected 0.97.

    I'll calculate the Pearson correlation coefficient between ad_spend and signups.

**Data:**
- ad_spend: 500, 800, 650, 1200, 900, 300, 1100, 700
- signups: 42, 55, 49, 71, 60, 35, 64, 58

**Calculations:**

Mean of ad_spend: (500+800+650+1200+900+300+1100+700)/8 = 6150/8 = 768.75

Mean of signups: (42+55+49+71+60+35+64+58)/8 = 434/8 = 54.25

**Deviations and products:**
| Week | ad_spend - 768.75 | signups - 54.25 | Product |
|------|------------------|-----------------|---------|
| 1 | -268.75 | -12.25 | 3291.19 |
| 2 | 31.25 | 0.75 | 23.44 |
| 3 | -118.75 | -5.25 | 623.44 |
| 4 | 431.25 | 16.75 | 7223.44 |
| 5 | 131.25 | 5.75 | 755.19 |
| 6 | -468.75 | -19.25 | 9023.44 |
| 7 | 331.25 | 9.75 | 3229.69 |
| 8 | -68.75 | 3.75 | -257.81 |

Sum of products: 23,891.98

**Variance calculations:**
- Variance of ad_spend: (72031.25 + 976.56 + 14101.56 + 185975.56 + 17226.56 + 219726.56 + 109726.56 + 4726.56) / 8 = 107995.31
- Variance of signups: (150.06 + 0.56 + 27.56 + 280.56 + 33.06 + 370.56 + 95.06 + 14.06) / 8 = 121.63

**Correlation:**
r = 23,891.98 / √(107,995.31 × 121.63) = 23,891.98 / √13,128.46 = 23,891.98 / 114.57 = 0.9875 ≈ **0.99**

Final answer: 0.99

    482 tokens in, 642 out · 5.0 s · $0.0037 · 1 message on Pro · answered by anthropic/claude-haiku-4.5 via Amazon Bedrock ·

How it compares on data analysis

Claude Haiku 4.5 beside the job's picks among every model. 16 models share the hard-problems pick, and Claude Haiku 4.5 isn't one of them.

Claude Haiku 4.5 beside other models on data analysis
ModelPassedOn ProCost per reply
GLM 5.3Our pick: hard problems (shared), best value5 of 5250 a month on Pro$0.0019
GPT-6 LunaOur pick: hard problems (shared), everyday5 of 560 a day on Pro$0.0002
Claude Haiku 4.53 of 5250 a month on Pro$0.0025
On Pro: Pro's count on each model. Cost: what OpenRouter charged us per reply, on average.

Every limit is published. Paid plans also have a monthly fair-use limit on AI cost: Plus $3.75, Pro $7.50, Max $20, Ultra $42, Studio $85. Using every message on your plan at typical sizes stays under it; very large messages and heavy research use it faster. Every limit, explained.

Claude Haiku 4.5 next to each model on data analysis

Claude Haiku 4.5 against each model on the same 5 data analysis prompts. One run each, through OpenRouter: a wait depends on the provider and the load that day, so a lead under 10% counts as close.

  • Against Claude Fable 5.1: Claude Haiku 4.5 answered 1.8× sooner (2.7 s to 4.9 s), cost 11.3× less, and passed 3 of 5 to its 5. Only Claude Fable 5.1 passed Average order value in August; Correlation between ad spend and sign-ups.

  • Against Claude Opus 5.5: Claude Haiku 4.5 answered 1.7× sooner (2.7 s to 4.6 s), cost 5.2× less, and passed 3 of 5 to its 5. Only Claude Opus 5.5 passed Average order value in August; Correlation between ad spend and sign-ups.

  • Against Claude Sonnet 5.5: Claude Haiku 4.5 answered 1.2× later (2.7 s to 2.3 s), cost 1.9× less, and passed 3 of 5 to its 5. Only Claude Sonnet 5.5 passed Average order value in August; Correlation between ad spend and sign-ups.

  • Against Claude Sonnet 5: Claude Haiku 4.5 answered 1.9× sooner (2.7 s to 5.3 s), cost 2.8× less, and passed 3 of 5 to its 4. Only Claude Sonnet 5 passed Average order value in August.

  • Against Claude Haiku 5.5: Claude Haiku 4.5 answered 1.2× sooner (2.7 s to 3.3 s), cost 7.0× more, and passed 3 of 5 to its 5. Only Claude Haiku 5.5 passed Average order value in August; Correlation between ad spend and sign-ups.

  • Against GPT-6 Astra: Claude Haiku 4.5 answered 1.2× sooner (2.7 s to 3.3 s), cost 6.2× less, and passed 3 of 5 to its 5. Only GPT-6 Astra passed Average order value in August; Correlation between ad spend and sign-ups.

  • Against GPT-6.1 Sol: Claude Haiku 4.5 answered within 10% of its time (2.7 s to 3.0 s), cost 1.6× more, and passed 3 of 5 to its 5. Only GPT-6.1 Sol passed Average order value in August; Correlation between ad spend and sign-ups.

  • Against GPT-6 Sol: Claude Haiku 4.5 answered within 10% of its time (2.7 s to 2.8 s), cost 1.4× less, and passed 3 of 5 to its 5. Only GPT-6 Sol passed Average order value in August; Correlation between ad spend and sign-ups.

  • Against GPT-6 Luna: Claude Haiku 4.5 answered within 10% of its time (2.7 s to 2.5 s), cost 14.2× more, and passed 3 of 5 to its 5. Only GPT-6 Luna passed Average order value in August; Correlation between ad spend and sign-ups.

  • Against Gemini 4 Argon: Claude Haiku 4.5 answered 2.5× sooner (2.7 s to 6.9 s), cost 1.1× less, and passed 3 of 5 to its 5. Only Gemini 4 Argon passed Average order value in August; Correlation between ad spend and sign-ups.

  • Against Gemini 3.1 Pro (preview): Claude Haiku 4.5 answered 3.2× sooner (2.7 s to 8.6 s), cost 6.2× less, and passed 3 of 5 to its 5. Only Gemini 3.1 Pro (preview) passed Average order value in August; Correlation between ad spend and sign-ups.

  • Against Gemini 3.8 Flash: Claude Haiku 4.5 answered 2.0× sooner (2.7 s to 5.4 s), cost 1.1× less, and passed 3 of 5 to its 4. Only Gemini 3.8 Flash passed Average order value in August.

  • Against DeepSeek V4.1 Flash: Claude Haiku 4.5 answered 3.2× later (2.7 s to 0.9 s), cost 2.9× more, and passed 3 of 5 to its 5. Only DeepSeek V4.1 Flash passed Average order value in August; Correlation between ad spend and sign-ups.

  • Against DeepSeek V4 Pro: Claude Haiku 4.5 answered 1.5× sooner (2.7 s to 4.1 s), cost 1.1× less, and passed 3 of 5 to its 5. Only DeepSeek V4 Pro passed Average order value in August; Correlation between ad spend and sign-ups.

  • Against Grok 4.7: Claude Haiku 4.5 answered 3.3× sooner (2.7 s to 8.9 s), cost 4.5× less, and passed 3 of 5 to its 5. Only Grok 4.7 passed Average order value in August; Correlation between ad spend and sign-ups.

  • Against Kimi K3: Claude Haiku 4.5 answered 1.3× sooner (2.7 s to 3.5 s), cost 2.7× less, and passed 3 of 5 to its 5. Only Kimi K3 passed Average order value in August; Correlation between ad spend and sign-ups.

  • Against GLM 5.3: Claude Haiku 4.5 answered 1.4× later (2.7 s to 1.9 s), cost 1.3× more, and passed 3 of 5 to its 5. Only GLM 5.3 passed Average order value in August; Correlation between ad spend and sign-ups.

  • Against GLM 5.3 Flash: Claude Haiku 4.5 answered 3.6× sooner (2.7 s to 9.8 s), cost 6.5× more, and passed 3 of 5 to its 4. Only GLM 5.3 Flash passed Average order value in August.

  • Against Mistral Large 4: Claude Haiku 4.5 answered 5.8× sooner (2.7 s to 15.8 s), cost 1.5× less, and passed 3 of 5 to its 5. Only Mistral Large 4 passed Average order value in August; Correlation between ad spend and sign-ups.

How these runs were done

Final answer. Automatic. The reply's last “Final answer:” line must hold the right value.

How the runs were done, and every data analysis prompt.

Claude Haiku 4.5, and data analysis, elsewhere

Questions

Is Claude Haiku 4.5 good for data analysis?

In our test runs it passed 3 of 5 data analysis prompts, 20th of the 20 models in llmwise. Every reply is on this page, so you can judge them yourself.

How many of my messages does a data analysis reply from Claude Haiku 4.5 use?

1 message each on Pro, where it has 250 a month on Pro. The price of a message is fixed and shown before you send it, however long the reply.

How were these runs done?

The same way for every model: each prompt sent through llmwise's own pipeline, each reply checked the same way. The methods page has every prompt and how each is scored.

Claude, GPT, Gemini, DeepSeek, Grok, Kimi, GLM, and Mistral, in one chat.

See what a message costs before you send it. Free is 5 messages to try; sign in with an email link, no password or card.

Start freeSee pricing