Tested prompt · Coding

Parse CSV with quoted fields: every AI model's reply, tested

We sent this hard coding prompt to all 20 models in llmwise, the same way the app sends a message, and checked every reply the same way. Here's each one as it came, with whether it passed, what it cost and how long it took.

Start free See pricing

Based on 20 of our test runs on October 15, 2026, through OpenRouter with the app's own prompt and settings. Updated October 15, 2026.

Short answer

19 of 20 models passed this coding prompt's check (tests); Claude Haiku 4.5 didn't. The cheapest reply that passed was GLM 5.3 Flash's, at $0.00026; the fastest, GLM 5.3's in 2.6 s. The dearest reply, Claude Fable 5.1's, cost 267 times as much ($0.0697).

The prompt, as sent, and its check

Checked by tests, the same way for every model.

Parse CSV with quoted fields (hard)

Write a JavaScript function parseCsv(text) that parses CSV text into an array of rows, each an array of strings.
- Fields are separated by commas, and rows by \n or \r\n.
- A field may be wrapped in double quotes. Inside quotes, commas and line breaks are part of the field, and "" stands for one double quote.
- A quote inside a field that doesn't start with one is an ordinary character.
- A line break at the very end of the text doesn't start a new row, and an empty string gives [].
Reply with the whole function in one ```javascript code block: plain JavaScript for Node.js 22, no imports, no TypeScript.

The reply's parseCsv runs against these 8 tests; it passes when all of them do.

parseCsv("a,b,c\n1,2,3")  // → [["a", "b", "c"], ["1", "2", "3"]]
parseCsv('"x, y",z')  // → [["x, y", "z"]]
parseCsv('"say ""hi""",2')  // → [["say \"hi\"", "2"]]
parseCsv('"line1\nline2",end\r\nnext,row\r\n')  // → [["line1\nline2", "end"], ["next", "row"]]
parseCsv("a,,c\n,\n")  // → [["a", "", "c"], ["", ""]]
parseCsv("")  // → []
parseCsv('he said "no",x')  // → [["he said \"no\"", "x"]]
parseCsv('a,"b"\n"",c')  // → [["a", "b"], ["", "c"]]

Exactly what this prompt's replies are checked against, with every other prompt of our test runs.

Every model's result

All 20 models on this prompt, in catalog order.

Every model's reply to “Parse CSV with quoted fields”
Model	Result	Cost	Time	Reply
Claude Fable 5.1Anthropic	Passed: All 8 tests passed.	$0.0697	14.3 s	1,229 tokens
Claude Opus 5.5Anthropic	Passed: All 8 tests passed.	$0.0273	11.9 s	563 tokens
Claude Sonnet 5.5Anthropic	Passed: All 8 tests passed.	$0.0114	5.3 s	978 tokens
Claude Sonnet 5Anthropic	Passed: All 8 tests passed.	$0.0187	15.8 s	463 tokens
Claude Haiku 5.5Anthropic	Passed: All 8 tests passed.	$0.00081	8.1 s	516 tokens
Claude Haiku 4.5Anthropic	Failed: 5 of 8 tests passed. First failure: parseCsv('"line1\nline2",end\r\nnext,row\r\n'): Expected values to be strictly deep-equal:	$0.0033	3.5 s	559 tokens
GPT-6 AstraOpenAI	Passed: All 8 tests passed.	$0.0314	9.1 s	313 tokens
GPT-6.1 SolOpenAI	Passed: All 8 tests passed.	$0.0038	8.4 s	321 tokens
GPT-6 SolOpenAI	Passed: All 8 tests passed.	$0.0081	8.9 s	349 tokens
GPT-6 LunaOpenAI	Passed: All 8 tests passed.	$0.00043	7.2 s	315 tokens
Gemini 4 ArgonGoogle	Passed: All 8 tests passed.	$0.0025	5.0 s	359 tokens
Gemini 3.1 Pro (preview)Google	Passed: All 8 tests passed.	$0.0291	20.6 s	462 tokens
Gemini 3.8 FlashGoogle	Passed: All 8 tests passed.	$0.0022	5.0 s	480 tokens
DeepSeek V4.1 FlashDeepSeek	Passed: All 8 tests passed.	$0.0042	9.1 s	399 tokens
DeepSeek V4 ProDeepSeek	Passed: All 8 tests passed.	$0.0357	122.3 s	325 tokens
Grok 4.7xAI	Passed: All 8 tests passed.	$0.0364	72.2 s	402 tokens
Kimi K3Moonshot	Passed: All 8 tests passed.	$0.0069	4.1 s	360 tokens
GLM 5.3Z.ai	Passed: All 8 tests passed.	$0.0024	2.6 s	456 tokens
GLM 5.3 FlashZ.ai	Passed: All 8 tests passed.	$0.00026	8.4 s	426 tokens
Mistral Large 4Mistral	Passed: All 8 tests passed.	$0.0118	57.3 s	697 tokens

Cost: what OpenRouter charged us for the reply. Time: from sending to the whole reply. Reply: its length in tokens, thinking not counted. In llmwise you pay per message, not per token: each of these replies counted as one message on Pro.

Every reply

The reply that failed first, then the ones that passed, from the cheapest up.

Claude Haiku 4.5

Failed: 5 of 8 tests passed. First failure: parseCsv('"line1\nline2",end\r\nnext,row\r\n'): Expected values to be strictly deep-equal:

```javascript
function parseCsv(text) {
  if (text === "") return [];

  const rows = [];
  let currentRow = [];
  let currentField = "";
  let inQuotes = false;
  let i = 0;

  while (i < text.length) {
    const char = text[i];

    if (inQuotes) {
      if (char === '"') {
        if (i + 1 < text.length && text[i + 1] === '"') {
          // "" inside quotes means a literal quote
          currentField += '"';
          i += 2;
        } else {
          // End of quoted field
          inQuotes = false;
          i++;
        }
      } else {
        // Any character (including comma, newline) inside quotes is part of the field
        currentField += char;
        i++;
      }
    } else {
      if (char === '"') {
        // Start of quoted field
        inQuotes = true;
        i++;
      } else if (char === ",") {
        // Field separator
        currentRow.push(currentField);
        currentField = "";
        i++;
      } else if (char === "\r") {
        // Handle \r\n or \r
        currentRow.push(currentField);
        currentField = "";
        if (i + 1 < text.length && text[i + 1] === "\n") {
          i += 2;
        } else {
          i++;
        }
        // Only add row if there's more content after the line break
        if (i < text.length) {
          rows.push(currentRow);
          currentRow = [];
        }
      } else if (char === "\n") {
        // Row separator
        currentRow.push(currentField);
        currentField = "";
        i++;
        // Only add row if there's more content after the line break
        if (i < text.length) {
          rows.push(currentRow);
          currentRow = [];
        }
      } else {
        // Ordinary character
        currentField += char;
        i++;
      }
    }
  }

  // Add the last field and row
  currentRow.push(currentField);
  rows.push(currentRow);

  return rows;
}
```

518 tokens in, 559 out · 3.5 s · $0.0033 · 1 message on Pro · answered by anthropic/claude-haiku-4.5 via Amazon Bedrock · September 27, 2026

GLM 5.3 Flash

Passed: All 8 tests passed.

```javascript
function parseCsv(text) {
  if (text === "") return [];
  const rows = [];
  let row = [];
  let field = "";
  let inQuotes = false;
  let i = 0;
  // Track whether the field started with a quote
  let fieldQuoted = false;

  function endField() {
    row.push(field);
    field = "";
    fieldQuoted = false;
  }

  function endRow() {
    endField();
    rows.push(row);
    row = [];
  }

  while (i < text.length) {
    const ch = text[i];

    if (inQuotes) {
      if (ch === '"') {
        if (text[i + 1] === '"') {
          field += '"';
          i += 2;
        } else {
          inQuotes = false;
          i++;
        }
      } else {
        field += ch;
        i++;
      }
      continue;
    }

    if (field === "" && !fieldQuoted && ch === '"') {
      fieldQuoted = true;
      inQuotes = true;
      i++;
      continue;
    }

    if (ch === ",") {
      endField();
      i++;
      continue;
    }

    if (ch === "\r" || ch === "\n") {
      // handle \r\n as one line break
      if (ch === "\r" && text[i + 1] === "\n") i++;
      // a line break at the very end of the text doesn't start a new row
      if (i + 1 >= text.length || (ch === "\r" && i + 2 >= text.length)) {
        i++;
        continue;
      }
      endRow();
      i++;
      continue;
    }

    field += ch;
    i++;
  }

  // final field/row if anything pending (a field was started)
  if (field !== "" || fieldQuoted || row.length > 0) {
    endRow();
  }

  return rows;
}
```

495 tokens in, 434 out (8 of them reasoning) · 8.4 s · $0.0003 · 1 message on Pro · answered by z-ai/glm-5.3-flash via AtlasCloud · September 27, 2026

GPT-6 Luna

Passed: All 8 tests passed.

```javascript
function parseCsv(text) {
  if (text === "") return [];

  const rows = [];
  let row = [];
  let field = "";
  let inQuotes = false;
  let endedWithRowBreak = false;

  for (let i = 0; i < text.length; i++) {
    const char = text[i];

    if (inQuotes) {
      if (char === '"') {
        if (text[i + 1] === '"') {
          field += '"';
          i++;
        } else {
          inQuotes = false;
        }
      } else {
        field += char;
      }
      endedWithRowBreak = false;
      continue;
    }

    if (char === '"' && field === "") {
      inQuotes = true;
      endedWithRowBreak = false;
    } else if (char === ",") {
      row.push(field);
      field = "";
      endedWithRowBreak = false;
    } else if (char === "\n" || (char === "\r" && text[i + 1] === "\n")) {
      row.push(field);
      rows.push(row);
      row = [];
      field = "";
      if (char === "\r") i++;
      endedWithRowBreak = true;
    } else {
      field += char;
      endedWithRowBreak = false;
    }
  }

  if (!endedWithRowBreak) {
    row.push(field);
    rows.push(row);
  }

  return rows;
}
```

481 tokens in, 762 out (447 of them reasoning) · 7.2 s · $0.0004 · 1 message on Pro · answered by openai/gpt-6-luna via OpenAI · September 27, 2026

Claude Haiku 5.5

Passed: All 8 tests passed.

```javascript
function parseCsv(text) {
  const rows = [];
  if (text === '') return rows;

  let row = [];
  let field = '';
  let inQuotes = false;
  let fieldStart = true; // true only at the very beginning of a field
  let i = 0;
  const n = text.length;

  while (i < n) {
    const c = text[i];

    if (inQuotes) {
      if (c === '"') {
        if (text[i + 1] === '"') {
          field += '"';
          i += 2;
        } else {
          inQuotes = false;
          i++;
        }
      } else {
        field += c;
        i++;
      }
      continue;
    }

    if (c === '"' && fieldStart) {
      inQuotes = true;
      fieldStart = false;
      i++;
      continue;
    }

    if (c === ',') {
      row.push(field);
      field = '';
      fieldStart = true;
      i++;
      continue;
    }

    if (c === '\n' || c === '\r') {
      row.push(field);
      rows.push(row);
      row = [];
      field = '';
      fieldStart = true;
      if (c === '\r' && text[i + 1] === '\n') i += 2;
      else i++;
      continue;
    }

    field += c;
    fieldStart = false;
    i++;
  }

  // Flush the last row unless the text ended with a line break
  if (row.length > 0 || field !== '' || !fieldStart) {
    row.push(field);
    rows.push(row);
  }

  return rows;
}
```

670 tokens in, 1,464 out (948 of them reasoning) · 8.1 s · $0.0008 · 1 message on Pro · answered by anthropic/claude-haiku-5.5 via Anthropic · October 7, 2026

Gemini 3.8 Flash

Passed: All 8 tests passed.

```javascript
function parseCsv(text) {
  if (text === '') return [];

  // Remove a single trailing newline if present
  if (text.endsWith('\r\n')) {
    text = text.slice(0, -2);
  } else if (text.endsWith('\n')) {
    text = text.slice(0, -1);
  }

  if (text === '') return [];

  const rows = [];
  let currentRow = [];
  let currentField = '';
  let inQuotes = false;
  let isStartOfField = true;

  for (let i = 0; i < text.length; i++) {
    const char = text[i];

    if (inQuotes) {
      if (char === '"') {
        if (text[i + 1] === '"') {
          currentField += '"';
          i++; // skip escaped quote
        } else {
          inQuotes = false;
        }
      } else {
        currentField += char;
      }
    } else {
      if (isStartOfField && char === '"') {
        inQuotes = true;
        isStartOfField = false;
      } else if (char === ',') {
        currentRow.push(currentField);
        currentField = '';
        isStartOfField = true;
      } else if (char === '\r' && text[i + 1] === '\n') {
        currentRow.push(currentField);
        rows.push(currentRow);
        currentRow = [];
        currentField = '';
        isStartOfField = true;
        i++; // skip '\n'
      } else if (char === '\n') {
        currentRow.push(currentField);
        rows.push(currentRow);
        currentRow = [];
        currentField = '';
        isStartOfField = true;
      } else {
        currentField += char;
        isStartOfField = false;
      }
    }
  }

  currentRow.push(currentField);
  rows.push(currentRow);

  return rows;
}
```

515 tokens in, 480 out · 5.0 s · $0.0022 · 1 message on Pro · answered by google/gemini-3.8-flash via Google · September 27, 2026

GLM 5.3

Passed: All 8 tests passed.

```javascript
function parseCsv(text) {
  if (text === '') return [];
  const rows = [];
  let row = [];
  let field = '';
  let inQuotes = false;
  let i = 0;
  const n = text.length;

  while (i < n) {
    const c = text[i];

    if (inQuotes) {
      if (c === '"') {
        if (text[i + 1] === '"') {
          field += '"';
          i += 2;
        } else {
          inQuotes = false;
          i++;
        }
      } else {
        field += c;
        i++;
      }
      continue;
    }

    if (c === '"' && field === '' && (i === 0 || /[\n\r,]/.test(text[i - 1]))) {
      inQuotes = true;
      i++;
    } else if (c === ',') {
      row.push(field);
      field = '';
      i++;
    } else if (c === '\r' && text[i + 1] === '\n') {
      row.push(field);
      rows.push(row);
      row = [];
      field = '';
      i += 2;
    } else if (c === '\n' || c === '\r') {
      row.push(field);
      rows.push(row);
      row = [];
      field = '';
      i++;
    } else {
      field += c;
      i++;
    }
  }

  // Only emit a final row if there's a pending field/row (avoids trailing newline creating empty row)
  if (field !== '' || row.length > 0) {
    row.push(field);
    rows.push(row);
  }

  return rows;
}
```

Notes on behavior:
- A quote only starts a quoted field at the beginning of a field (start of input, after a comma, or after a line break), per your third rule; otherwise `"` is literal.
- `""` inside quotes becomes a single `"`.
- Embedded `\n`/`\r\n` inside quotes stays in the field; a trailing newline at the end of input doesn't create an empty final row; `''` returns `[]`.

518 tokens in, 463 out (7 of them reasoning) · 2.6 s · $0.0024 · 1 message on Pro · answered by z-ai/glm-5.3 via Wafer · September 27, 2026

Gemini 4 Argon

Passed: All 8 tests passed.

```javascript
function parseCsv(text) {
  if (text === "") return [];

  const rows = [];
  let row = [];
  let field = "";
  let inQuotes = false;
  let fieldStart = true;

  for (let i = 0; i < text.length; i++) {
    const ch = text[i];

    if (inQuotes) {
      if (ch === '"') {
        if (text[i + 1] === '"') {
          field += '"';
          i++;
        } else {
          inQuotes = false;
        }
      } else {
        field += ch;
      }
      continue;
    }

    if (fieldStart && ch === '"') {
      inQuotes = true;
      fieldStart = false;
    } else if (ch === ",") {
      row.push(field);
      field = "";
      fieldStart = true;
    } else if (ch === "\n" || (ch === "\r" && text[i + 1] === "\n")) {
      if (ch === "\r") i++;
      row.push(field);
      rows.push(row);
      row = [];
      field = "";
      fieldStart = true;
      if (i === text.length - 1) return rows;
    } else {
      field += ch;
      fieldStart = false;
    }
  }

  row.push(field);
  rows.push(row);
  return rows;
}
```

736 tokens in, 359 out · 5.0 s · $0.0025 · 1 message on Pro · answered by google/gemini-4-argon via Google AI Studio · October 15, 2026

GPT-6.1 Sol

Passed: All 8 tests passed.

```javascript
function parseCsv(text) {
  if (text === "") return [];

  const rows = [];
  let row = [];
  let field = "";
  let inQuotes = false;
  let atFieldStart = true;
  let rowStarted = false;

  for (let i = 0; i < text.length; i++) {
    const char = text[i];
    rowStarted = true;

    if (inQuotes) {
      if (char === '"') {
        if (text[i + 1] === '"') {
          field += '"';
          i++;
        } else {
          inQuotes = false;
        }
      } else {
        field += char;
      }
    } else if (char === '"' && atFieldStart) {
      inQuotes = true;
      atFieldStart = false;
    } else if (char === ",") {
      row.push(field);
      field = "";
      atFieldStart = true;
    } else if (
      char === "\n" ||
      (char === "\r" && text[i + 1] === "\n")
    ) {
      if (char === "\r") i++;
      row.push(field);
      rows.push(row);
      row = [];
      field = "";
      atFieldStart = true;
      rowStarted = false;
    } else {
      field += char;
      atFieldStart = false;
    }
  }

  if (rowStarted) {
    row.push(field);
    rows.push(row);
  }

  return rows;
}
```

483 tokens in, 665 out (344 of them reasoning) · 8.4 s · $0.0038 · 1 message on Pro · answered by openai/gpt-6.1-sol via OpenAI · September 29, 2026

DeepSeek V4.1 Flash

Passed: All 8 tests passed.

```javascript
function parseCsv(text) {
  const rows = [];
  let row = [];
  let field = '';
  let inQuotes = false;
  let pending = false; // true if there is an in-progress field/row
  let i = 0;
  const n = text.length;

  while (i < n) {
    const ch = text[i];

    if (inQuotes) {
      if (ch === '"') {
        if (text[i + 1] === '"') {
          field += '"';
          i += 2;
        } else {
          inQuotes = false;
          i += 1;
        }
      } else {
        field += ch;
        i += 1;
      }
      continue;
    }

    if (ch === '"' && field === '') {
      // A quote only opens a quoted field when it is the first char of the field.
      inQuotes = true;
      pending = true;
      i += 1;
      continue;
    }

    if (ch === ',') {
      row.push(field);
      field = '';
      pending = true;
      i += 1;
      continue;
    }

    if (ch === '\n' || ch === '\r') {
      if (ch === '\r' && text[i + 1] === '\n') i += 1; // treat CRLF as one break
      row.push(field);
      rows.push(row);
      row = [];
      field = '';
      pending = false;
      i += 1;
      continue;
    }

    field += ch;
    pending = true;
    i += 1;
  }

  // A line break at the very end leaves "pending" false, so no empty trailing row.
  if (pending) {
    row.push(field);
    rows.push(row);
  }

  return rows;
}
```

522 tokens in, 3,473 out (3,074 of them reasoning) · 9.1 s · $0.0042 · 1 message on Pro · answered by deepseek/deepseek-v4.1-flash via Together · October 8, 2026

Kimi K3

Passed: All 8 tests passed.

```javascript
function parseCsv(text) {
  if (text === "") return [];

  const rows = [];
  let row = [];
  let field = "";
  let inQuotes = false;
  let quoted = false; // current field started with a quote
  let afterQuote = false; // just closed a quoted field

  for (let i = 0; i < text.length; i++) {
    const c = text[i];

    if (inQuotes) {
      if (c === '"') {
        if (text[i + 1] === '"') {
          field += '"';
          i++;
        } else {
          inQuotes = false;
          afterQuote = true;
        }
      } else {
        field += c;
      }
    } else if (c === '"' && field === "" && !afterQuote) {
      inQuotes = true;
      quoted = true;
    } else if (c === ",") {
      row.push(field);
      field = "";
      quoted = false;
      afterQuote = false;
    } else if (c === "\n" || c === "\r") {
      if (c === "\r" && text[i + 1] === "\n") i++;
      row.push(field);
      rows.push(row);
      row = [];
      field = "";
      quoted = false;
      afterQuote = false;
    } else {
      field += c;
    }
  }

  // Trailing line break doesn't start a new row; otherwise flush last field/row.
  if (field !== "" || quoted || row.length > 0) {
    row.push(field);
    rows.push(row);
  }

  return rows;
}
```

568 tokens in, 367 out (7 of them reasoning) · 4.1 s · $0.0069 · 1 message on Pro · answered by moonshotai/kimi-k3 via Together · September 27, 2026

GPT-6 Sol

Passed: All 8 tests passed.

```javascript
function parseCsv(text) {
  if (text === "") return [];

  const rows = [];
  let row = [];
  let field = "";
  let inQuotes = false;
  let atFieldStart = true;
  let endedWithRowBreak = false;

  for (let i = 0; i < text.length; i++) {
    const char = text[i];

    if (inQuotes) {
      if (char === '"') {
        if (text[i + 1] === '"') {
          field += '"';
          i++;
        } else {
          inQuotes = false;
        }
      } else {
        field += char;
      }
      endedWithRowBreak = false;
    } else if (char === '"' && atFieldStart) {
      inQuotes = true;
      atFieldStart = false;
      endedWithRowBreak = false;
    } else if (char === ",") {
      row.push(field);
      field = "";
      atFieldStart = true;
      endedWithRowBreak = false;
    } else if (char === "\n" || (char === "\r" && text[i + 1] === "\n")) {
      row.push(field);
      rows.push(row);
      row = [];
      field = "";
      atFieldStart = true;
      endedWithRowBreak = true;
      if (char === "\r") i++;
    } else {
      field += char;
      atFieldStart = false;
      endedWithRowBreak = false;
    }
  }

  if (!endedWithRowBreak) {
    row.push(field);
    rows.push(row);
  }

  return rows;
}
```

481 tokens in, 712 out (363 of them reasoning) · 8.9 s · $0.0081 · 1 message on Pro · answered by openai/gpt-6-sol via OpenAI · September 27, 2026

Claude Sonnet 5.5

Passed: All 8 tests passed.

```javascript
function parseCsv(text) {
  const rows = [];
  let row = [];
  let field = "";
  let inQuotes = false;
  let fieldStarted = false; // any content or quote seen in current row
  const n = text.length;

  for (let i = 0; i < n; i++) {
    const c = text[i];

    if (inQuotes) {
      if (c === '"') {
        if (text[i + 1] === '"') {
          field += '"';
          i++;
        } else {
          inQuotes = false;
        }
      } else {
        field += c;
      }
      continue;
    }

    if (c === '"' && field === "" && !fieldStarted) {
      inQuotes = true;
      fieldStarted = true;
    } else if (c === ",") {
      row.push(field);
      field = "";
      fieldStarted = false;
      rowHasContent = true;
    } else if (c === "\n" || (c === "\r" && text[i + 1] === "\n")) {
      if (c === "\r") i++;
      row.push(field);
      rows.push(row);
      row = [];
      field = "";
      fieldStarted = false;
      rowHasContent = false;
    } else {
      field += c;
      fieldStarted = true;
    }
  }

  // Final row: only if the text didn't end with a line break
  if (field !== "" || fieldStarted || row.length > 0 || inQuotes) {
    row.push(field);
    rows.push(row);
  }

  return rows;
}
```

Wait — `rowHasContent` is undeclared, which would throw in strict mode. Here is the corrected version:

```javascript
function parseCsv(text) {
  const rows = [];
  let row = [];
  let field = "";
  let inQuotes = false;
  let fieldStarted = false;
  const n = text.length;

  for (let i = 0; i < n; i++) {
    const c = text[i];

    if (inQuotes) {
      if (c === '"') {
        if (text[i + 1] === '"') {
          field += '"';
          i++;
        } else {
          inQuotes = false;
        }
      } else {
        field += c;
      }
      continue;
    }

    if (c === '"' && !fieldStarted) {
      inQuotes = true;
      fieldStarted = true;
    } else if (c === ",") {
      row.push(field);
      field = "";
      fieldStarted = false;
    } else if (c === "\n" || (c === "\r" && text[i + 1] === "\n")) {
      if (c === "\r") i++;
      row.push(field);
      rows.push(row);
      row = [];
      field = "";
      fieldStarted = false;
    } else {
      field += c;
      fieldStarted = true;
    }
  }

  if (fieldStarted || field !== "" || row.length > 0 || inQuotes) {
    row.push(field);
    rows.push(row);
  }

  return rows;
}
```

671 tokens in, 978 out · 5.3 s · $0.0114 · 1 message on Pro · answered by anthropic/claude-sonnet-5.5 via Anthropic · September 28, 2026

Mistral Large 4

Passed: All 8 tests passed.

```javascript
function parseCsv(text) {
  if (text === '') return [];

  const rows = [];
  let row = [];
  let field = '';
  let inQuotes = false;
  let hasField = false; // current field has begun (quote opened or char seen)
  let i = 0;

  const endField = () => {
    row.push(field);
    field = '';
    hasField = false;
  };
  const endRow = () => {
    endField();
    rows.push(row);
    row = [];
  };

  while (i < text.length) {
    const ch = text[i];

    if (inQuotes) {
      if (ch === '"') {
        if (text[i + 1] === '"') {
          field += '"';
          i += 2;
        } else {
          inQuotes = false;
          i++;
        }
      } else {
        field += ch;
        i++;
      }
      continue;
    }

    if (ch === '"' && !hasField) {
      inQuotes = true;
      hasField = true;
      i++;
    } else if (ch === ',') {
      endField();
      i++;
    } else if (ch === '\n' || ch === '\r') {
      if (ch === '\r' && text[i + 1] === '\n') i++;
      endRow();
      i++;
    } else {
      field += ch;
      hasField = true;
      i++;
    }
  }

  // Push the final row unless the text ended with a line break.
  if (hasField || row.length > 0) {
    endField();
    rows.push(row);
  }

  return rows;
}
```

491 tokens in, 5,505 out (4,808 of them reasoning) · 57.3 s · $0.0118 · 1 message on Pro · answered by mistralai/mistral-large-4-0 via Mistral · October 9, 2026

Claude Sonnet 5

Passed: All 8 tests passed.

```javascript
function parseCsv(text) {
  const rows = [];
  let row = [];
  let field = '';
  let i = 0;
  const n = text.length;
  let inQuotes = false;

  while (i < n) {
    const c = text[i];

    if (inQuotes) {
      if (c === '"') {
        if (text[i + 1] === '"') {
          field += '"';
          i += 2;
        } else {
          inQuotes = false;
          i++;
        }
      } else {
        field += c;
        i++;
      }
      continue;
    }

    if (c === '"' && field === '') {
      inQuotes = true;
      i++;
    } else if (c === ',') {
      row.push(field);
      field = '';
      i++;
    } else if (c === '\r' && text[i + 1] === '\n') {
      row.push(field);
      rows.push(row);
      row = [];
      field = '';
      i += 2;
    } else if (c === '\n') {
      row.push(field);
      rows.push(row);
      row = [];
      field = '';
      i++;
    } else {
      field += c;
      i++;
    }
  }

  // Only push the final row if there's actually pending content;
  // a trailing line break should not create an extra empty row.
  if (field !== '' || row.length > 0) {
    row.push(field);
    rows.push(row);
  }

  return rows;
}
```

667 tokens in, 1,737 out (1,274 of them reasoning) · 15.8 s · $0.0187 · 1 message on Pro · answered by anthropic/claude-sonnet-5 via Claude Platform on AWS · September 27, 2026

Claude Opus 5.5

Passed: All 8 tests passed.

```javascript
function parseCsv(text) {
  const rows = [];
  const n = text.length;
  if (n === 0) return rows;

  let row = [];
  let field = '';
  let inQuotes = false;
  let atFieldStart = true; // no characters consumed yet for the current field
  let rowPending = false;  // current row has consumed at least one character
  let i = 0;

  while (i < n) {
    const c = text[i];

    if (inQuotes) {
      if (c === '"') {
        if (text[i + 1] === '"') {
          field += '"';
          i += 2;
        } else {
          inQuotes = false;
          i += 1;
        }
      } else {
        field += c;
        i += 1;
      }
      continue;
    }

    if (c === '"' && atFieldStart) {
      inQuotes = true;
      atFieldStart = false;
      rowPending = true;
      i += 1;
      continue;
    }

    if (c === ',') {
      row.push(field);
      field = '';
      atFieldStart = true;
      rowPending = true;
      i += 1;
      continue;
    }

    if (c === '\n' || (c === '\r' && text[i + 1] === '\n')) {
      row.push(field);
      rows.push(row);
      row = [];
      field = '';
      atFieldStart = true;
      rowPending = false;
      i += c === '\r' ? 2 : 1;
      continue;
    }

    field += c;
    atFieldStart = false;
    rowPending = true;
    i += 1;
  }

  if (rowPending || inQuotes) {
    row.push(field);
    rows.push(row);
  }

  return rows;
}
```

669 tokens in, 1,200 out (637 of them reasoning) · 11.9 s · $0.0273 · 1 message on Pro · answered by anthropic/claude-opus-5.5 via Claude Platform on AWS · September 27, 2026

Gemini 3.1 Pro

Passed: All 8 tests passed.

```javascript
function parseCsv(text) {
  if (text === "") return [];
  
  const rows = [];
  let currentRow = [];
  let currentField = "";
  let inQuotes = false;
  
  for (let i = 0; i < text.length; i++) {
    const c = text[i];
    
    if (inQuotes) {
      if (c === '"') {
        if (i + 1 < text.length && text[i + 1] === '"') {
          currentField += '"';
          i++; // Skip escaped quote
        } else {
          inQuotes = false;
        }
      } else {
        currentField += c;
      }
    } else {
      if (c === '"' && currentField === "") {
        inQuotes = true;
      } else if (c === ',') {
        currentRow.push(currentField);
        currentField = "";
      } else if (c === '\r' && i + 1 < text.length && text[i + 1] === '\n') {
        currentRow.push(currentField);
        rows.push(currentRow);
        currentRow = [];
        currentField = "";
        i++; // Skip \n
      } else if (c === '\n') {
        currentRow.push(currentField);
        rows.push(currentRow);
        currentRow = [];
        currentField = "";
      } else {
        currentField += c;
      }
    }
  }
  
  // If the text didn't end with a newline, push the final field and row.
  // We also push if the very last character was a comma (meaning an empty field at the end).
  if (currentField !== "" || currentRow.length > 0 || (text.length > 0 && text[text.length - 1] === ',')) {
    currentRow.push(currentField);
    rows.push(currentRow);
  }
  
  return rows;
}
```

515 tokens in, 2,339 out (1,877 of them reasoning) · 20.6 s · $0.0291 · 1 message on Pro · answered by google/gemini-3.1-pro-preview via Google · September 27, 2026

GPT-6 Astra

Passed: All 8 tests passed.

```javascript
function parseCsv(text) {
  if (text === "") return [];

  const rows = [];
  let row = [];
  let field = "";
  let inQuotes = false;
  let atFieldStart = true;

  for (let i = 0; i < text.length; i++) {
    const char = text[i];

    if (inQuotes) {
      if (char === '"') {
        if (text[i + 1] === '"') {
          field += '"';
          i++;
        } else {
          inQuotes = false;
        }
      } else {
        field += char;
      }
      continue;
    }

    if (char === '"' && atFieldStart) {
      inQuotes = true;
      atFieldStart = false;
    } else if (char === ",") {
      row.push(field);
      field = "";
      atFieldStart = true;
    } else if (
      char === "\n" ||
      (char === "\r" && text[i + 1] === "\n")
    ) {
      if (char === "\r") i++;
      row.push(field);
      rows.push(row);
      row = [];
      field = "";
      atFieldStart = true;
    } else {
      field += char;
      atFieldStart = false;
    }
  }

  if (row.length > 0 || !atFieldStart) {
    row.push(field);
    rows.push(row);
  }

  return rows;
}
```

481 tokens in, 532 out (219 of them reasoning) · 9.1 s · $0.0314 · 1 message on Pro · answered by openai/gpt-6-astra via OpenAI · September 27, 2026

DeepSeek V4 Pro

Passed: All 8 tests passed.

```javascript
function parseCsv(text) {
  const rows = [];
  let row = [];
  let field = '';
  let fieldStarted = false;
  let inQuotes = false;

  for (let i = 0; i < text.length; i++) {
    const ch = text[i];

    if (inQuotes) {
      if (ch === '"') {
        if (text[i + 1] === '"') {
          field += '"';
          i++;
        } else {
          inQuotes = false;
        }
      } else {
        field += ch;
      }
      continue;
    }

    if (ch === '"' && !fieldStarted) {
      inQuotes = true;
      fieldStarted = true;
    } else if (ch === ',') {
      row.push(field);
      field = '';
      fieldStarted = false;
    } else if (ch === '\n') {
      row.push(field);
      field = '';
      fieldStarted = false;
      rows.push(row);
      row = [];
    } else if (ch === '\r') {
      row.push(field);
      field = '';
      fieldStarted = false;
      rows.push(row);
      row = [];
      if (text[i + 1] === '\n') {
        i++;
      }
    } else {
      field += ch;
      fieldStarted = true;
    }
  }

  if (fieldStarted || row.length > 0) {
    row.push(field);
    rows.push(row);
  }

  return rows;
}
```

494 tokens in, 8,886 out (8,561 of them reasoning) · 122.3 s · $0.0357 · 1 message on Pro · answered by deepseek/deepseek-v4-pro-0813 via Wafer · October 8, 2026

Grok 4.7

Passed: All 8 tests passed.

```javascript
function parseCsv(text) {
  if (text === "") return [];

  const rows = [];
  let row = [];
  let field = "";
  let inQuotes = false;
  let atFieldStart = true;
  let endedOnBreak = false;

  for (let i = 0; i < text.length; ) {
    const c = text[i];

    if (inQuotes) {
      if (c === '"') {
        if (text[i + 1] === '"') {
          field += '"';
          i += 2;
        } else {
          inQuotes = false;
          i += 1;
        }
      } else {
        field += c;
        i += 1;
      }
      endedOnBreak = false;
      continue;
    }

    if (atFieldStart && c === '"') {
      inQuotes = true;
      atFieldStart = false;
      endedOnBreak = false;
      i += 1;
      continue;
    }

    if (c === ",") {
      row.push(field);
      field = "";
      atFieldStart = true;
      endedOnBreak = false;
      i += 1;
      continue;
    }

    if (c === "\n" || (c === "\r" && text[i + 1] === "\n")) {
      row.push(field);
      rows.push(row);
      row = [];
      field = "";
      atFieldStart = true;
      endedOnBreak = true;
      i += c === "\r" ? 2 : 1;
      continue;
    }

    field += c;
    atFieldStart = false;
    endedOnBreak = false;
    i += 1;
  }

  if (!endedOnBreak) {
    row.push(field);
    rows.push(row);
  }

  return rows;
}
```

1,718 tokens in, 5,854 out (5,452 of them reasoning) · 72.2 s · $0.0364 · 1 message on Pro · answered by x-ai/grok-4.7 via xAI · October 2, 2026

Claude Fable 5.1

Passed: All 8 tests passed.

```javascript
function parseCsv(text) {
  const rows = [];
  if (text.length === 0) return rows;

  let row = [];
  let field = '';
  let i = 0;
  const n = text.length;

  while (i < n) {
    const ch = text[i];

    if (field === '' && ch === '"' && fieldStart) {
      // Quoted field
      i++;
      while (i < n) {
        const c = text[i];
        if (c === '"') {
          if (text[i + 1] === '"') {
            field += '"';
            i += 2;
          } else {
            i++;
            break;
          }
        } else {
          field += c;
          i++;
        }
      }
      // After closing quote, consume any trailing chars until delimiter/newline
      while (i < n && text[i] !== ',' && text[i] !== '\n' && text[i] !== '\r') {
        field += text[i];
        i++;
      }
      fieldStart = false;
      continue;
    }

    if (ch === ',') {
      row.push(field);
      field = '';
      fieldStart = true;
      i++;
    } else if (ch === '\n' || (ch === '\r' && text[i + 1] === '\n')) {
      row.push(field);
      rows.push(row);
      row = [];
      field = '';
      fieldStart = true;
      i += ch === '\r' ? 2 : 1;
    } else {
      field += ch;
      fieldStart = false;
      i++;
    }
  }

  // Handle final field/row unless text ended exactly on a line break
  const endsWithNewline = text.endsWith('\n');
  if (!endsWithNewline) {
    row.push(field);
    rows.push(row);
  }

  return rows;

  // fieldStart tracks whether we're at the beginning of a field (so a quote
  // there opens a quoted field, while a quote mid-field is a literal char).
  var fieldStart;
}
```

Note: `fieldStart` is declared with `var` (hoisted) and starts `undefined`, which is falsy — so the first field wouldn't be treated as quoted. Here is the corrected version with it initialized properly:

```javascript
function parseCsv(text) {
  const rows = [];
  if (text.length === 0) return rows;

  let row = [];
  let field = '';
  let fieldStart = true;
  let i = 0;
  const n = text.length;

  while (i < n) {
    const ch = text[i];

    if (fieldStart && ch === '"') {
      i++;
      while (i < n) {
        const c = text[i];
        if (c === '"') {
          if (text[i + 1] === '"') {
            field += '"';
            i += 2;
          } else {
            i++;
            break;
          }
        } else {
          field += c;
          i++;
        }
      }
      while (i < n && text[i] !== ',' && text[i] !== '\n' && text[i] !== '\r') {
        field += text[i];
        i++;
      }
      fieldStart = false;
      continue;
    }

    if (ch === ',') {
      row.push(field);
      field = '';
      fieldStart = true;
      i++;
    } else if (ch === '\n' || (ch === '\r' && text[i + 1] === '\n')) {
      row.push(field);
      rows.push(row);
      row = [];
      field = '';
      fieldStart = true;
      i += ch === '\r' ? 2 : 1;
    } else {
      field += ch;
      fieldStart = false;
      i++;
    }
  }

  if (!text.endsWith('\n')) {
    row.push(field);
    rows.push(row);
  }

  return rows;
}
```

Use the second block.

669 tokens in, 1,229 out · 14.3 s · $0.0697 · 1 message on Pro · answered by anthropic/claude-fable-5.1 via Anthropic · September 27, 2026

More coding prompts

The other coding prompts, each with every model's reply, and the results across all five.

Questions

Which AI does best on “Parse CSV with quoted fields”?

What does a reply to “Parse CSV with quoted fields” cost?

Through the models' APIs, what OpenRouter charged us ran from $0.00026 (GLM 5.3 Flash) to $0.0697 (Claude Fable 5.1) for this prompt. In llmwise you don't pay by the token: a reply like these counts as one message on Pro, whichever model answers.

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 free See pricing