Claude vs ChatGPT for math

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.

Math: Claude Haiku 5.5 passed 5 of 5 and GPT-6.1 Sol 5 of 5. Their median waits were close, 1.7 s against 1.8 s. Claude Haiku 5.5 cost 5.9× less, $0.0007 against $0.0043 for the 5 replies. Claude Haiku 5.5's replies ran 122% longer, in tokens of reply, thinking not counted.

1. 1. A discount, then sales tax

   Everyday. Claude: 6 of 6 models passed, Claude Haiku 5.5 the cheapest at $0.0001 · GPT: 4 of 4 models passed, GPT-6 Luna the cheapest at $0.0001. The prompt in full.

   Claude Haiku 5.5 and GPT-6.1 Sol both passed this one. Claude Haiku 5.5 answered sooner (1.3 s to 1.5 s), and Claude Haiku 5.5's reply cost less ($0.0001 to $0.0006); Claude Haiku 5.5 wrote 61 tokens of reply, GPT-6.1 Sol 39.

   • Claude Haiku 5.5

     Passed: Final answer 66: right.

     Discounted price: $80 × 0.75 = $60
     Sales tax: $60 × 0.10 = $6
     Total: $60 + $6 = $66
     
     Final answer: $66

     544 tokens in, 61 out · 1.3 s · $0.0001 · 1 message on Pro · answered by anthropic/claude-haiku-5.5 via Anthropic · October 7, 2026

   • GPT-6.1 Sol

     Passed: Final answer 66: right.

     Discounted price: $80 × 0.75 = $60  
     After sales tax: $60 × 1.10 = $66
     
     Final answer: $66

     402 tokens in, 39 out · 1.5 s · $0.0006 · 1 message on Pro · answered by openai/gpt-6.1-sol via OpenAI · September 29, 2026

2. 2. Pens at 3 for $4

   Everyday. Claude: 6 of 6 models passed, Claude Haiku 5.5 the cheapest at $0.0002 · GPT: 3 of 4 models passed, GPT-6.1 Sol the cheapest at $0.0010. The prompt in full.

   Claude Haiku 5.5 and GPT-6.1 Sol both passed this one. Claude Haiku 5.5 answered sooner (1.7 s to 2.4 s), and Claude Haiku 5.5's reply cost less ($0.0002 to $0.0010); Claude Haiku 5.5 wrote 212 tokens of reply, GPT-6.1 Sol 89.

   • Claude Haiku 5.5

     Passed: Final answer Buy 3 packs of 3 and 1 single pen, costing 13.50: right.

     Options per pen:
     - Bundle of 3 costs $4 (about $1.33 per pen).
     - Single pen costs $1.50.
     
     For 10 pens, use as many bundles as possible:
     - 3 bundles = 9 pens = $12.00
     - 1 single pen = $1.50
     - Total = $13.50
     
     Check alternatives:
     - 2 bundles + 4 singles = $8 + $6 = $14.00
     - 0 bundles + 10 singles = $15.00
     - 3 bundles + 1 single is the cheapest.
     
     Final answer: Buy 3 packs of 3 and 1 single pen, costing $13.50

     548 tokens in, 212 out · 1.7 s · $0.0002 · 1 message on Pro · answered by anthropic/claude-haiku-5.5 via Anthropic · October 7, 2026

   • GPT-6.1 Sol

     Passed: Final answer Buy three 3-pen bundles and one single pen for 13.50.: right.

     Three singles cost $4.50, so each 3-for-$4 bundle saves $0.50. Use as many bundles as possible: 3 bundles give 9 pens, then buy 1 single.
     
     Cost: \(3 \times \$4 + \$1.50 = \$13.50\).
     
     Final answer: Buy three 3-pen bundles and one single pen for $13.50.

     404 tokens in, 127 out (38 of them reasoning) · 2.4 s · $0.0010 · 1 message on Pro · answered by openai/gpt-6.1-sol via OpenAI · September 29, 2026

3. 3. Compound interest over three years

   Everyday. Claude: 6 of 6 models passed, Claude Haiku 5.5 the cheapest at $0.0001 · GPT: 4 of 4 models passed, GPT-6 Luna the cheapest at $0.0001. The prompt in full.

   Claude Haiku 5.5 and GPT-6.1 Sol both passed this one. Claude Haiku 5.5 answered sooner (1.3 s to 1.7 s), and Claude Haiku 5.5's reply cost less ($0.0001 to $0.0007); Claude Haiku 5.5 wrote 104 tokens of reply, GPT-6.1 Sol 64.

   • Claude Haiku 5.5

     Passed: Final answer 315.25: right.

     Compound interest formula: A = P(1 + r)^t
     
     A = 2000 × (1.05)^3
     (1.05)^3 = 1.157625
     A = 2000 × 1.157625 = 2315.25
     
     Interest earned = 2315.25 − 2000 = 315.25
     
     Final answer: $315.25

     542 tokens in, 104 out · 1.3 s · $0.0001 · 1 message on Pro · answered by anthropic/claude-haiku-5.5 via Anthropic · October 7, 2026

   • GPT-6.1 Sol

     Passed: Final answer 315.25: right.

     After 3 years, the balance is:
     \[
     2000(1.05)^3 = \$2,315.25
     \]
     
     Interest earned:
     \[
     \$2,315.25 - \$2,000 = \$315.25
     \]
     
     Final answer: $315.25

     404 tokens in, 64 out · 1.7 s · $0.0007 · 1 message on Pro · answered by openai/gpt-6.1-sol via OpenAI · September 29, 2026

4. 4. Four-digit numbers whose digits sum to 9

   Hard. Claude: 6 of 6 models passed, Claude Haiku 5.5 the cheapest at $0.0002 · GPT: 4 of 4 models passed, GPT-6 Luna the cheapest at $0.0001. The prompt in full.

   Claude Haiku 5.5 and GPT-6.1 Sol both passed this one. Claude Haiku 5.5 answered sooner (1.9 s to 2.7 s), and Claude Haiku 5.5's reply cost less ($0.0002 to $0.0011); Claude Haiku 5.5 wrote 305 tokens of reply, GPT-6.1 Sol 108.

   • Claude Haiku 5.5

     Passed: Final answer 165: right.

     We need four-digit numbers d1 d2 d3 d4 with d1 from 1 to 9, d2, d3, d4 from 0 to 9, and digit sum 9.
     
     Substitute e1 = d1 − 1, so e1 ranges 0 to 8, and e1 + d2 + d3 + d4 = 8 with all variables from 0 to 9.
     
     Since the sum is 8, no variable can exceed 8, so the upper bound of 9 is never violated. The number of nonnegative solutions is C(8+3, 3) = C(11, 3) = 165.
     
     Check: total solutions with d1 ≥ 0 allowed (i.e., first digit could be 0) for sum 9 across four digits is C(12, 3) = 220. Those with d1 = 0 correspond to three digits summing to 9: C(11, 2) = 55. So 220 − 55 = 165, which matches.
     
     Final answer: 165

     513 tokens in, 305 out · 1.9 s · $0.0002 · 1 message on Pro · answered by anthropic/claude-haiku-5.5 via Anthropic · October 7, 2026

   • GPT-6.1 Sol

     Passed: Final answer 165: right.

     Let the digits be \(a,b,c,d\), with \(a\ge1\). Set \(x=a-1\). Then
     \[
     x+b+c+d=8,
     \]
     where all four variables are nonnegative. Each resulting digit is automatically at most 9.
     
     By stars and bars, the number of solutions is
     \[
     \binom{8+4-1}{4-1}=\binom{11}{3}=165.
     \]
     
     Final answer: 165

     380 tokens in, 147 out (39 of them reasoning) · 2.7 s · $0.0011 · 1 message on Pro · answered by openai/gpt-6.1-sol via OpenAI · September 29, 2026

5. 5. The highest of three dice is a 5

   Hard. Claude: 6 of 6 models passed, Claude Haiku 5.5 the cheapest at $0.0002 · GPT: 4 of 4 models passed, GPT-6 Luna the cheapest at $0.0001. The prompt in full.

   Claude Haiku 5.5 and GPT-6.1 Sol both passed this one. GPT-6.1 Sol answered sooner (1.8 s to 2.1 s), and Claude Haiku 5.5's reply cost less ($0.0002 to $0.0008); Claude Haiku 5.5 wrote 179 tokens of reply, GPT-6.1 Sol 87.

   • Claude Haiku 5.5

     Passed: Final answer 61/216: right.

     The highest number is exactly 5 when all three dice show at most 5, minus the cases where all three show at most 4:
     
     - All dice ≤ 5: 5³ = 125 outcomes
     - All dice ≤ 4: 4³ = 64 outcomes
     - Exactly max 5: 125 − 64 = 61 outcomes
     
     Out of 6³ = 216 total outcomes, the probability is 61/216. Since 61 is prime and doesn't divide 216, the fraction is already in lowest terms.
     
     Final answer: 61/216

     540 tokens in, 250 out (71 of them reasoning) · 2.1 s · $0.0002 · 1 message on Pro · answered by anthropic/claude-haiku-5.5 via Anthropic · October 7, 2026

   • GPT-6.1 Sol

     Passed: Final answer (61/216 ): right.

     The highest number is exactly 5 if all three dice show at most 5, but not all show at most 4.
     
     \[
     P(\max=5)=\left(\frac56\right)^3-\left(\frac46\right)^3
     =\frac{125-64}{216}=\frac{61}{216}.
     \]
     
     Final answer: \(\frac{61}{216}\)

     395 tokens in, 87 out · 1.8 s · $0.0008 · 1 message on Pro · answered by openai/gpt-6.1-sol via OpenAI · September 29, 2026

• Claude Fable 5.1 vs GPT-6 Astra: 5 of 5 each; GPT-6 Astra answered 1.8× sooner at the median and cost 1.4× less. Claude Fable 5.1 vs GPT-6 Astra, on every job.

• Claude Fable 5.1 vs GPT-6.1 Sol: 5 of 5 each; GPT-6.1 Sol answered 2.6× sooner at the median and cost 13.5× less.

• Claude Fable 5.1 vs GPT-6 Sol: 5 of 5 each; GPT-6 Sol answered 2.1× sooner at the median and cost 6.4× less.

• Claude Fable 5.1 vs GPT-6 Luna: Claude Fable 5.1 5 of 5, GPT-6 Luna 4; GPT-6 Luna answered 2.1× sooner at the median and cost 118.2× less.

• Claude Opus 5.5 vs GPT-6 Astra: 5 of 5 each; GPT-6 Astra answered 1.5× sooner at the median and Claude Opus 5.5 cost 1.4× less. Claude Opus 5.5 vs GPT-6 Astra, on every job.

• Claude Opus 5.5 vs GPT-6.1 Sol: 5 of 5 each; GPT-6.1 Sol answered 2.2× sooner at the median and cost 6.9× less.

• Claude Opus 5.5 vs GPT-6 Sol: 5 of 5 each; GPT-6 Sol answered 1.8× sooner at the median and cost 3.3× less. Claude Opus 5.5 vs GPT-6 Sol, on every job.

• Claude Opus 5.5 vs GPT-6 Luna: Claude Opus 5.5 5 of 5, GPT-6 Luna 4; GPT-6 Luna answered 1.8× sooner at the median and cost 60.5× less.

• Claude Sonnet 5.5 vs GPT-6 Astra: 5 of 5 each; Claude Sonnet 5.5 answered 1.6× sooner at the median and cost 3.5× less.

• Claude Sonnet 5.5 vs GPT-6.1 Sol: 5 of 5 each; they took about as long and GPT-6.1 Sol cost 2.9× less. GPT-6.1 Sol vs Claude Sonnet 5.5, on every job.

• Claude Sonnet 5.5 vs GPT-6 Sol: 5 of 5 each; Claude Sonnet 5.5 answered 1.3× sooner at the median and GPT-6 Sol cost 1.4× less. Claude Sonnet 5.5 vs GPT-6 Sol, on every job.

• Claude Sonnet 5.5 vs GPT-6 Luna: Claude Sonnet 5.5 5 of 5, GPT-6 Luna 4; Claude Sonnet 5.5 answered 1.3× sooner at the median and GPT-6 Luna cost 25.1× less.

• Claude Sonnet 5 vs GPT-6 Astra: 5 of 5 each; GPT-6 Astra answered 1.5× sooner at the median and Claude Sonnet 5 cost 2.6× less.

• Claude Sonnet 5 vs GPT-6.1 Sol: 5 of 5 each; GPT-6.1 Sol answered 2.2× sooner at the median and cost 3.8× less.

• Claude Sonnet 5 vs GPT-6 Sol: 5 of 5 each; GPT-6 Sol answered 1.8× sooner at the median and cost 1.8× less. Claude Sonnet 5 vs GPT-6 Sol, on every job.

• Claude Sonnet 5 vs GPT-6 Luna: Claude Sonnet 5 5 of 5, GPT-6 Luna 4; GPT-6 Luna answered 1.8× sooner at the median and cost 33.5× less.

• Claude Haiku 4.5 vs GPT-6 Astra: 5 of 5 each; Claude Haiku 4.5 answered 1.2× sooner at the median and cost 5.6× less.

• Claude Haiku 4.5 vs GPT-6.1 Sol: 5 of 5 each; GPT-6.1 Sol answered 1.3× sooner at the median and cost 1.8× less.

• Claude Haiku 4.5 vs GPT-6 Sol: 5 of 5 each; they took about as long and Claude Haiku 4.5 cost 1.2× less. Claude Haiku 4.5 vs GPT-6 Sol, on every job.

• Claude Haiku 4.5 vs GPT-6 Luna: Claude Haiku 4.5 5 of 5, GPT-6 Luna 4; they took about as long and GPT-6 Luna cost 15.6× less. Claude Haiku 4.5 vs GPT-6 Luna, on every job.

• Claude Haiku 5.5 vs GPT-6 Astra: 5 of 5 each; Claude Haiku 5.5 answered 1.5× sooner at the median and cost 58.1× less.

• Claude Haiku 5.5 vs GPT-6.1 Sol: 5 of 5 each; they took about as long and Claude Haiku 5.5 cost 5.9× less.

• Claude Haiku 5.5 vs GPT-6 Sol: 5 of 5 each; Claude Haiku 5.5 answered 1.3× sooner at the median and cost 12.3× less.

• Claude Haiku 5.5 vs GPT-6 Luna: Claude Haiku 5.5 5 of 5, GPT-6 Luna 4; Claude Haiku 5.5 answered 1.3× sooner at the median and GPT-6 Luna cost 1.5× less. Claude Haiku 5.5 vs GPT-6 Luna, on every job.

Every Claude and GPT reply above was checked the same way as every other model's, by the rules published with the prompts: how the math prompts are scored, and each one in full.

• The lineups

  Claude: 6 models, Claude Fable 5.1, Claude Opus 5.5, Claude Sonnet 5.5, Claude Sonnet 5, Claude Haiku 4.5, and Claude Haiku 5.5. GPT: 4 models, GPT-6 Astra, GPT-6.1 Sol, GPT-6 Sol, and GPT-6 Luna.

• Price per message

  The least expensive Claude model is Claude Haiku 5.5 (60 messages a day on Pro); the least expensive GPT model is GPT-6 Luna (60 messages a day on Pro).

• Context window

  Claude goes up to 1M tokens (Claude Fable 5.1); GPT up to 1.05M tokens (GPT-6 Astra).

• Images and PDFs

  Every model here reads images. Every model here takes a PDF as a whole file.

• On the Free plan

  Free's one-time trial of 5 messages covers Claude Haiku 5.5. Paid plans have them, with messages every month: Claude Fable 5.1 and GPT-6 Astra from Pro up.

In llmwise, a message to Claude or GPT goes to the model's maker, or through OpenRouter when llmwise can't reach the maker directly. The Privacy Policy has the details.