Model vs model

Claude Opus 5.5 vs Claude Haiku 4.5

Claude Opus 5.5 and Claude Haiku 4.5 are both in llmwise. What each gets on every plan, what it reads, how it's served and what happens when its provider fails, from the catalog and the code that runs them.

Start free See pricing

Model prices and specs checked September 28, 2026 against OpenRouter's Claude Opus 5.5 page, OpenRouter's Claude Haiku 4.5 page. Updated September 28, 2026.

Short answer

Claude Haiku 4.5 gets four times as many messages: up to 250 messages a month on Pro, against up to 62 messages a month for Claude Opus 5.5. Otherwise, only Claude Opus 5.5 can be passed to another model by its maker's safety system and only Claude Haiku 4.5 is in the free trial. In our test runs, Claude Opus 5.5 passed 49 of the 50 prompts both answered and Claude Haiku 4.5 43; 8 prompts split them, most on coding (5 to 3).

Claude Opus 5.5 vs Claude Haiku 4.5, prompt by prompt

Every prompt Claude Opus 5.5 and Claude Haiku 4.5 both answered, compared directly, their biggest differences first. One run each, through OpenRouter: a wait depends on the provider and the load that day, so a lead under 10% counts as close.

Of the 50 prompts both answered, both passed 42, only Claude Opus 5.5 passed 7, only Claude Haiku 4.5 passed 1, and neither passed 0. Claude Opus 5.5 answered sooner on 0 of the 50 and Claude Haiku 4.5 on 47; the rest were within 10% of each other. The 50 replies cost $0.5360 on Claude Opus 5.5 and $0.0741 on Claude Haiku 4.5: 7.2× less on Claude Haiku 4.5.

The 8 prompts only one of Claude Opus 5.5 and Claude Haiku 4.5 passed

Evaluate an arithmetic expression, no eval (coding): Claude Opus 5.5 passed and Claude Haiku 4.5 didn't. Claude Opus 5.5: All 15 tests passed. Claude Haiku 4.5: 14 of 15 tests passed. First failure: evaluate("-2 ^ 2"): Expected values to be strictly deep-equal:
Parse CSV with quoted fields (coding): Claude Opus 5.5 passed and Claude Haiku 4.5 didn't. Claude Opus 5.5: All 8 tests passed. Claude Haiku 4.5: 5 of 8 tests passed. First failure: parseCsv('"line1\nline2",end\r\nnext,row\r\n'): Expected values to be strictly deep-equal:
A product announcement with five rules (writing): Claude Opus 5.5 passed and Claude Haiku 4.5 didn't. Claude Opus 5.5: Graded 4.7 of 5 on average (lowest 4). Claude Haiku 4.5: Graded 3.7 of 5 on average (lowest 3).
Argue both sides of free buses (writing): Claude Haiku 4.5 passed and Claude Opus 5.5 didn't. Claude Opus 5.5: Graded 4.7 of 5 on average (lowest 4); but a paragraph of 93 words, over the 90 allowed. Claude Haiku 4.5: Graded 4.0 of 5 on average (lowest 4).
Average order value in August (data analysis): Claude Opus 5.5 passed and Claude Haiku 4.5 didn't. Claude Opus 5.5: Final answer 300.00: right. Claude Haiku 4.5: Final answer 275.00; expected 300.
Correlation between ad spend and sign-ups (data analysis): Claude Opus 5.5 passed and Claude Haiku 4.5 didn't. Claude Opus 5.5: Final answer 0.97: right. Claude Haiku 4.5: Final answer 0.99; expected 0.97.
A frustrated customer (customer support): Claude Opus 5.5 passed and Claude Haiku 4.5 didn't. Claude Opus 5.5: Graded 4.0 of 5 on average (lowest 3). Claude Haiku 4.5: Graded 3.3 of 5 on average (lowest 3).
A message with a planted instruction (customer support): Claude Opus 5.5 passed and Claude Haiku 4.5 didn't. Claude Opus 5.5: Graded 4.3 of 5 on average (lowest 3). Claude Haiku 4.5: Graded 4.3 of 5 on average (lowest 3); but 165 words, over the 150 allowed.

Job by job, the widest gaps first

Customer support: Claude Opus 5.5 passed 5 of 5 and Claude Haiku 4.5 3 of 5. Claude Haiku 4.5 answered 2.4× sooner at the median, 6.4 s against 2.7 s. Claude Haiku 4.5 cost 7.4× less, $0.0540 against $0.0073 for the 5 replies. Claude Opus 5.5's replies ran 27% longer, in tokens of reply, thinking not counted.
Coding: Claude Opus 5.5 passed 5 of 5 and Claude Haiku 4.5 3 of 5. Claude Haiku 4.5 answered 3.1× sooner at the median, 7.2 s against 2.3 s. Claude Haiku 4.5 cost 6.2× less, $0.0853 against $0.0137 for the 5 replies. Their replies ran to about the same length.
Data analysis: Claude Opus 5.5 passed 5 of 5 and Claude Haiku 4.5 3 of 5. Claude Haiku 4.5 answered 1.7× sooner at the median, 4.6 s against 2.7 s. Claude Haiku 4.5 cost 5.2× less, $0.0654 against $0.0125 for the 5 replies. Claude Haiku 4.5's replies ran 76% longer, in tokens of reply, thinking not counted.
Writing: Claude Opus 5.5 passed 4 of 5 and Claude Haiku 4.5 4 of 5. Claude Haiku 4.5 answered 3.7× sooner at the median, 8.2 s against 2.2 s. Claude Haiku 4.5 cost 16.3× less, $0.0923 against $0.0057 for the 5 replies. Claude Opus 5.5's replies ran 55% longer, in tokens of reply, thinking not counted.
Summarization: Claude Opus 5.5 passed 5 of 5 and Claude Haiku 4.5 5 of 5. Claude Haiku 4.5 answered 2.2× sooner at the median, 3.7 s against 1.7 s. Claude Haiku 4.5 cost 9.3× less, $0.0475 against $0.0051 for the 5 replies. Claude Opus 5.5's replies ran 74% longer, in tokens of reply, thinking not counted.
SQL: Claude Opus 5.5 passed 5 of 5 and Claude Haiku 4.5 5 of 5. Claude Haiku 4.5 answered 3.5× sooner at the median, 4.3 s against 1.2 s. Claude Haiku 4.5 cost 9.3× less, $0.0459 against $0.0049 for the 5 replies. Claude Opus 5.5's replies ran 114% longer, in tokens of reply, thinking not counted.
Translation: Claude Opus 5.5 passed 5 of 5 and Claude Haiku 4.5 5 of 5. Claude Haiku 4.5 answered 3.0× sooner at the median, 5.5 s against 1.8 s. Claude Haiku 4.5 cost 7.3× less, $0.0540 against $0.0074 for the 5 replies. Claude Opus 5.5's replies ran 11% longer, in tokens of reply, thinking not counted.
RAG and answering from documents: Claude Opus 5.5 passed 5 of 5 and Claude Haiku 4.5 5 of 5. Claude Haiku 4.5 answered 2.4× sooner at the median, 2.9 s against 1.2 s. Claude Haiku 4.5 cost 7.3× less, $0.0367 against $0.0050 for the 5 replies. Claude Opus 5.5's replies ran 29% longer, in tokens of reply, thinking not counted.
Agents and tool use: Claude Opus 5.5 passed 5 of 5 and Claude Haiku 4.5 5 of 5. Claude Haiku 4.5 answered 2.8× sooner at the median, 3.1 s against 1.1 s. Claude Haiku 4.5 cost 5.3× less, $0.0252 against $0.0047 for the 5 replies. Claude Haiku 4.5's replies ran 26% longer, in tokens of reply, thinking not counted.
Math: Claude Opus 5.5 passed 5 of 5 and Claude Haiku 4.5 5 of 5. Claude Haiku 4.5 answered 1.7× sooner at the median, 3.9 s against 2.2 s. Claude Haiku 4.5 cost 3.9× less, $0.0298 against $0.0077 for the 5 replies. Claude Haiku 4.5's replies ran 64% longer, in tokens of reply, thinking not counted.

All 50 prompts: who passed, who answered sooner, who cost less

Claude Opus 5.5 and Claude Haiku 4.5 on each prompt of our test runs
Prompt	Result	Sooner	Cheaper
Turn a title into a URL slug	Both passed	Claude Haiku 4.5, 2.0×took 3.1 s and 1.5 s	Claude Haiku 4.5, 5.8×cost $0.0055 and $0.0010
Parse a duration like “1h 30m”	Both passed	Claude Haiku 4.5, 3.1×took 7.2 s and 2.3 s	Claude Haiku 4.5, 5.2×cost $0.0112 and $0.0022
Merge overlapping intervals	Both passed	Claude Haiku 4.5, 1.9×took 3.7 s and 2.0 s	Claude Haiku 4.5, 4.6×cost $0.0077 and $0.0017
Evaluate an arithmetic expression, no eval	Only Claude Opus 5.5	Claude Haiku 4.5, 2.1×took 15.8 s and 7.6 s	Claude Haiku 4.5, 6.0×cost $0.0336 and $0.0056
Parse CSV with quoted fields	Only Claude Opus 5.5	Claude Haiku 4.5, 3.4×took 11.9 s and 3.5 s	Claude Haiku 4.5, 8.2×cost $0.0273 and $0.0033
Announce a second bakery shop on LinkedIn	Both passed	Claude Haiku 4.5, 3.4×took 8.7 s and 2.6 s	Claude Haiku 4.5, 13.8×cost $0.0175 and $0.0013
Rewrite corporate jargon in plain words	Both passed	Claude Haiku 4.5, 3.1×took 5.1 s and 1.6 s	Claude Haiku 4.5, 9.5×cost $0.0078 and $0.0008
Decline a meeting and offer two times	Both passed	Claude Haiku 4.5, 2.0×took 4.1 s and 2.0 s	Claude Haiku 4.5, 5.7×cost $0.0056 and $0.0010
A product announcement with five rules	Only Claude Opus 5.5	Claude Haiku 4.5, 3.7×took 8.2 s and 2.2 s	Claude Haiku 4.5, 14.1×cost $0.0170 and $0.0012
Argue both sides of free buses	Only Claude Haiku 4.5	Claude Haiku 4.5, 6.5×took 22.5 s and 3.4 s	Claude Haiku 4.5, 31.9×cost $0.0443 and $0.0014
A discount, then sales tax	Both passed	Claude Haiku 4.5, 2.4×took 2.7 s and 1.1 s	Claude Haiku 4.5, 4.9×cost $0.0037 and $0.0007
Pens at 3 for $4	Both passed	Claude Haiku 4.5, 2.5×took 5.3 s and 2.2 s	Claude Haiku 4.5, 5.9×cost $0.0084 and $0.0014
Compound interest over three years	Both passed	Closetook 3.7 s and 3.5 s	Claude Haiku 4.5, 3.3×cost $0.0043 and $0.0013
Four-digit numbers whose digits sum to 9	Both passed	Claude Haiku 4.5, 1.7×took 4.7 s and 2.9 s	Claude Haiku 4.5, 3.2×cost $0.0077 and $0.0024
The highest of three dice is a 5	Both passed	Claude Haiku 4.5, 1.7×took 3.9 s and 2.2 s	Claude Haiku 4.5, 3.2×cost $0.0057 and $0.0018
An article in three bullets	Both passed	Claude Haiku 4.5, 4.5×took 8.1 s and 1.8 s	Claude Haiku 4.5, 15.9×cost $0.0171 and $0.0011
An email thread in one sentence	Both passed	Claude Haiku 4.5, 2.0×took 2.6 s and 1.3 s	Claude Haiku 4.5, 6.1×cost $0.0044 and $0.0007
Decisions and action items from a meeting	Both passed	Claude Haiku 4.5, 3.5×took 5.8 s and 1.7 s	Claude Haiku 4.5, 11.1×cost $0.0118 and $0.0011
A quarterly memo for the CEO	Both passed	Claude Haiku 4.5, 2.3×took 3.5 s and 1.5 s	Claude Haiku 4.5, 6.2×cost $0.0075 and $0.0012
A study with a negative result	Both passed	Closetook 3.7 s and 3.4 s	Claude Haiku 4.5, 6.6×cost $0.0066 and $0.0010
The region with the most revenue	Both passed	Claude Haiku 4.5, 1.5×took 4.1 s and 2.7 s	Claude Haiku 4.5, 3.3×cost $0.0087 and $0.0026
Average order value in August	Only Claude Opus 5.5	Claude Haiku 4.5, 2.1×took 4.4 s and 2.1 s	Claude Haiku 4.5, 4.6×cost $0.0107 and $0.0023
Revenue change from July to August	Both passed	Claude Haiku 4.5, 1.2×took 5.1 s and 4.2 s	Claude Haiku 4.5, 5.4×cost $0.0120 and $0.0022
A median, filtered two ways	Both passed	Claude Haiku 4.5, 2.3×took 4.6 s and 2.0 s	Claude Haiku 4.5, 4.8×cost $0.0079 and $0.0016
Correlation between ad spend and sign-ups	Only Claude Opus 5.5	Claude Haiku 4.5, 2.3×took 11.5 s and 5.0 s	Claude Haiku 4.5, 7.1×cost $0.0261 and $0.0037
A late order	Both passed	Claude Haiku 4.5, 3.9×took 7.8 s and 2.0 s	Claude Haiku 4.5, 8.4×cost $0.0114 and $0.0014
A return inside the window	Both passed	Claude Haiku 4.5, 1.9×took 4.4 s and 2.4 s	Claude Haiku 4.5, 5.6×cost $0.0074 and $0.0013
A frustrated customer	Only Claude Opus 5.5	Claude Haiku 4.5, 2.2×took 6.4 s and 3.0 s	Claude Haiku 4.5, 7.1×cost $0.0111 and $0.0016
A refund request outside the window	Both passed	Claude Haiku 4.5, 2.2×took 6.1 s and 2.7 s	Claude Haiku 4.5, 8.2×cost $0.0106 and $0.0013
A message with a planted instruction	Only Claude Opus 5.5	Claude Haiku 4.5, 2.5×took 7.9 s and 3.2 s	Claude Haiku 4.5, 7.6×cost $0.0136 and $0.0018
A delivery message into Spanish	Both passed	Claude Haiku 4.5, 2.1×took 3.2 s and 1.5 s	Claude Haiku 4.5, 5.6×cost $0.0064 and $0.0011
A product description into French	Both passed	Claude Haiku 4.5, 4.1×took 5.8 s and 1.4 s	Claude Haiku 4.5, 9.9×cost $0.0113 and $0.0011
A meeting note into German	Both passed	Claude Haiku 4.5, 3.0×took 5.5 s and 1.8 s	Claude Haiku 4.5, 9.3×cost $0.0096 and $0.0010
Idioms into natural Japanese	Both passed	Closetook 3.6 s and 3.3 s	Claude Haiku 4.5, 3.3×cost $0.0062 and $0.0019
A lease clause into Brazilian Portuguese	Both passed	Claude Haiku 4.5, 3.2×took 11.4 s and 3.5 s	Claude Haiku 4.5, 9.4×cost $0.0204 and $0.0022
Customers in one country	Both passed	Claude Haiku 4.5, 2.2×took 2.5 s and 1.1 s	Claude Haiku 4.5, 7.4×cost $0.0048 and $0.0007
Count orders by status	Both passed	Claude Haiku 4.5, 3.0×took 2.5 s and 0.8 s	Claude Haiku 4.5, 7.7×cost $0.0049 and $0.0006
Revenue by category	Both passed	Claude Haiku 4.5, 3.5×took 4.3 s and 1.2 s	Claude Haiku 4.5, 8.7×cost $0.0092 and $0.0011
Every customer, even those without orders	Both passed	Claude Haiku 4.5, 2.4×took 5.0 s and 2.1 s	Claude Haiku 4.5, 9.3×cost $0.0094 and $0.0010
Monthly revenue with a running total	Both passed	Claude Haiku 4.5, 4.8×took 7.8 s and 1.6 s	Claude Haiku 4.5, 11.2×cost $0.0176 and $0.0016
A fact from one section	Both passed	Claude Haiku 4.5, 2.4×took 2.5 s and 1.1 s	Claude Haiku 4.5, 5.9×cost $0.0054 and $0.0009
Core hours and start times	Both passed	Claude Haiku 4.5, 2.3×took 2.6 s and 1.1 s	Claude Haiku 4.5, 5.9×cost $0.0054 and $0.0009
Two sections in one answer	Both passed	Claude Haiku 4.5, 2.2×took 2.9 s and 1.3 s	Claude Haiku 4.5, 5.0×cost $0.0053 and $0.0011
A later amendment changes the answer	Both passed	Claude Haiku 4.5, 4.0×took 7.7 s and 1.9 s	Claude Haiku 4.5, 12.0×cost $0.0148 and $0.0012
A question the handbook doesn't answer	Both passed	Claude Haiku 4.5, 2.8×took 3.4 s and 1.2 s	Claude Haiku 4.5, 6.2×cost $0.0057 and $0.0009
Pick the tool and work out the date	Both passed	Claude Haiku 4.5, 2.5×took 2.7 s and 1.1 s	Claude Haiku 4.5, 4.9×cost $0.0041 and $0.0008
Convert a currency	Both passed	Claude Haiku 4.5, 2.8×took 2.8 s and 1.0 s	Claude Haiku 4.5, 4.9×cost $0.0041 and $0.0008
Book a meeting from a sentence	Both passed	Claude Haiku 4.5, 5.9×took 7.2 s and 1.2 s	Claude Haiku 4.5, 5.6×cost $0.0059 and $0.0010
Search, but don't book	Both passed	Claude Haiku 4.5, 2.8×took 3.1 s and 1.1 s	Claude Haiku 4.5, 4.9×cost $0.0046 and $0.0010
Two calls with a unit conversion	Both passed	Claude Haiku 4.5, 2.8×took 3.4 s and 1.2 s	Claude Haiku 4.5, 6.1×cost $0.0064 and $0.0011

Claude Opus 5.5 vs Claude Haiku 4.5 in our test runs

Claude Opus 5.5 and Claude Haiku 4.5 on the same prompts, job by job: how many replies passed their check.

Based on 100 of our test runs on September 27, 2026, through OpenRouter with the app's own prompt and settings. Updated September 28, 2026.

Claude Opus 5.5 and Claude Haiku 4.5 in our test runs, job by job
Job	Claude Opus 5.5	Claude Haiku 4.5
Coding	5 of 5	3 of 5
Writing	4 of 5	4 of 5
Math	5 of 5	5 of 5
Summarization	5 of 5	5 of 5
Data analysis	5 of 5	3 of 5
Customer support	5 of 5	3 of 5
Translation	5 of 5	5 of 5
SQL	5 of 5	5 of 5
RAG and answering from documents	5 of 5	5 of 5
Agents and tool use	5 of 5	5 of 5
All jobs	49 of 50 · 1 message each · $0.0107 a reply	43 of 50 · 1 message each · $0.0015 a reply

The same prompts, sent to both through llmwise's own pipeline and checked the same way: five for each job. A bold count is the one that passed more. Messages: what each reply counted as on Pro. Cost: what OpenRouter charged us per reply, on average. Every prompt, and how it's scored.

Where they differed most, coding: both replies to a prompt only one of them passed.

A hard prompt: “Evaluate an arithmetic expression, no eval”, in full.

Claude Opus 5.5

Passed: All 15 tests passed.

```javascript
function evaluate(expression) {
  if (typeof expression !== "string") {
    throw new Error("Expression must be a string");
  }

  // Tokenizer
  const tokens = [];
  const numberRe = /\d+(?:\.\d+)?|\.\d+/y;
  let i = 0;
  while (i < expression.length) {
    const ch = expression[i];
    if (/\s/.test(ch)) {
      i++;
      continue;
    }
    if ("+-*/^()".includes(ch)) {
      tokens.push({ type: ch });
      i++;
      continue;
    }
    numberRe.lastIndex = i;
    const m = numberRe.exec(expression);
    if (m) {
      tokens.push({ type: "num", value: parseFloat(m[0]) });
      i += m[0].length;
      continue;
    }
    throw new Error(`Unexpected character '${ch}' at position ${i}`);
  }

  // Recursive descent parser
  let pos = 0;
  const peek = () => tokens[pos];
  const accept = (type) => {
    if (tokens[pos] && tokens[pos].type === type) {
      pos++;
      return true;
    }
    return false;
  };

  // expr := term (('+' | '-') term)*
  function parseExpr() {
    let value = parseTerm();
    for (;;) {
      if (accept("+")) value += parseTerm();
      else if (accept("-")) value -= parseTerm();
      else return value;
    }
  }

  // term := unary (('*' | '/') unary)*
  function parseTerm() {
    let value = parseUnary();
    for (;;) {
      if (accept("*")) value *= parseUnary();
      else if (accept("/")) value /= parseUnary();
      else return value;
    }
  }

  // unary := '-' unary | power
  function parseUnary() {
    if (accept("-")) return -parseUnary();
    return parsePower();
  }

  // power := primary ('^' unary)?   (right-associative; exponent may be negated)
  function parsePower() {
    const base = parsePrimary();
    if (accept("^")) {
      const exponent = parseUnary();
      return Math.pow(base, exponent);
    }
    return base;
  }

  // primary := number | '(' expr ')'
  function parsePrimary() {
    const tok = peek();
    if (!tok) throw new Error("Unexpected end of expression");
    if (tok.type === "num") {
      pos++;
      return tok.value;
    }
    if (accept("(")) {
      const value = parseExpr();
      if (!accept(")")) throw new Error("Expected ')'");
      return value;
    }
    throw new Error(`Unexpected token '${tok.type}'`);
  }

  if (tokens.length === 0) throw new Error("Empty expression");
  const result = parseExpr();
  if (pos !== tokens.length) {
    const tok = tokens[pos];
    throw new Error(
      `Unexpected token '${tok.type === "num" ? tok.value : tok.type}'`
    );
  }
  return result;
}
```

762 tokens in, 1,491 out (435 of them reasoning) · 15.8 s · $0.0336 · 1 message on Pro · answered by anthropic/claude-opus-5.5 via Claude Platform on AWS · September 27, 2026

Claude Haiku 4.5

Failed: 14 of 15 tests passed. First failure: evaluate("-2 ^ 2"): Expected values to be strictly deep-equal:

```javascript
function evaluate(expression) {
  const tokens = tokenize(expression);
  const parser = new Parser(tokens);
  const result = parser.parseExpression();
  
  if (parser.pos !== tokens.length) {
    throw new Error("Unexpected tokens after expression");
  }
  
  return result;
}

function tokenize(expression) {
  const tokens = [];
  let i = 0;
  
  while (i < expression.length) {
    const ch = expression[i];
    
    if (/\s/.test(ch)) {
      i++;
      continue;
    }
    
    if (/\d/.test(ch)) {
      let num = '';
      while (i < expression.length && /[\d.]/.test(expression[i])) {
        num += expression[i];
        i++;
      }
      const parsed = parseFloat(num);
      if (isNaN(parsed)) {
        throw new Error(`Invalid number: ${num}`);
      }
      tokens.push({ type: 'NUMBER', value: parsed });
      continue;
    }
    
    if ('+-*/^()'.includes(ch)) {
      tokens.push({ type: ch, value: ch });
      i++;
      continue;
    }
    
    throw new Error(`Unexpected character: ${ch}`);
  }
  
  return tokens;
}

class Parser {
  constructor(tokens) {
    this.tokens = tokens;
    this.pos = 0;
  }
  
  peek() {
    return this.tokens[this.pos];
  }
  
  consume() {
    return this.tokens[this.pos++];
  }
  
  parseExpression() {
    return this.parseAddSub();
  }
  
  parseAddSub() {
    let left = this.parseMulDiv();
    
    while (this.peek() && (this.peek().type === '+' || this.peek().type === '-')) {
      const op = this.consume().type;
      const right = this.parseMulDiv();
      left = op === '+' ? left + right : left - right;
    }
    
    return left;
  }
  
  parseMulDiv() {
    let left = this.parsePower();
    
    while (this.peek() && (this.peek().type === '*' || this.peek().type === '/')) {
      const op = this.consume().type;
      const right = this.parsePower();
      left = op === '*' ? left * right : left / right;
    }
    
    return left;
  }
  
  parsePower() {
    let left = this.parseUnary();
    
    if (this.peek() && this.peek().type === '^') {
      this.consume();
      const right = this.parsePower();
      left = Math.pow(left, right);
    }
    
    return left;
  }
  
  parseUnary() {
    if (this.peek() && this.peek().type === '-') {
      this.consume();
      return -this.parseUnary();
    }
    
    if (this.peek() && this.peek().type === '+') {
      this.consume();
      return this.parseUnary();
    }
    
    return this.parsePrimary();
  }
  
  parsePrimary() {
    const token = this.peek();
    
    if (!token) {
      throw new Error("Unexpected end of expression");
    }
    
    if (token.type === 'NUMBER') {
      this.consume();
      return token.value;
    }
    
    if (token.type === '(') {
      this.consume();
      const result = this.parseExpression();
      
      if (!this.peek() || this.peek().type !== ')') {
        throw new Error("Missing closing parenthesis");
      }
      
      this.consume();
      return result;
    }
    
    throw new Error(`Unexpected token: ${token.value}`);
  }
}
```

599 tokens in, 1,007 out · 7.6 s · $0.0056 · 1 message on Pro · answered by anthropic/claude-haiku-4.5 via Amazon Bedrock · September 27, 2026

Claude Opus 5.5 and Claude Haiku 4.5 on every plan

Whether the one-time free trial reaches each model, then each paid plan's messages on it.

Messages on Claude Opus 5.5 and Claude Haiku 4.5, plan by plan
Plan	Price	Claude Opus 5.5	Claude Haiku 4.5
Free	$0	Not in the trial	In the one-time trial of 5 messages
Pro	$20 a month	Up to 62 a month	Up to 250 a month
Max	$50 a month	Up to 200 a month	Up to 800 a month
Ultra	$100 a month	Up to 450 a month	Up to 1,800 a month
Studio	$200 a month	Up to 1,000 a month	Up to 4,000 a month

Prices don't include tax, which is added where it applies and shown before you pay. A paid plan's month is one allowance shared by every model, so each monthly count is the most you get if all of it goes to that model. It renews each billing period; everyday models refill daily at 00:00 UTC. Long chats count more per reply. How pricing works.

Every limit is published. Paid plans also have a monthly fair-use limit on AI cost: 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.

What differs

Messages on Pro
Claude Haiku 4.5 gets four times as many messages: up to 250 messages a month on Pro, against up to 62 messages a month for Claude Opus 5.5.
Context window
Claude Opus 5.5 takes up to 1M tokens; Claude Haiku 4.5 up to 200K tokens. A chat in llmwise holds up to 200k tokens, which fits in either, so the difference shows only through each maker's own API.
Images and PDFs
Both read images. Both take a PDF as the whole file, pages and all.
On Free
Claude Haiku 4.5 is in the free trial; Claude Opus 5.5 needs a paid plan.
Where messages go
Both are Claude models, served the same way. Sent to Anthropic directly.

Fact by fact

Claude Opus 5.5 and Claude Haiku 4.5, fact by fact
Fact	Claude Opus 5.5	Claude Haiku 4.5
Context window	1M tokens	200K tokens
Reads images	Yes	Yes
PDFs	Whole file	Whole file
Reasoning	Yes	No
API price (September 2026)	$4.00 in / $20.00 out per million tokens	$1.00 in / $5.00 out per million tokens
A typical message at API prices (4,000 tokens in, 700 out)	$0.0300	$0.0075
A $10 top-up adds	50 messages	200 messages
Where a message goes	Sent to Anthropic directly.	Sent to Anthropic directly.
If the provider fails	If Anthropic fails before the reply starts (an overload, a server error, a dropped connection), llmwise sends the same request to Claude Opus 5.5 through OpenRouter instead.	If Anthropic fails before the reply starts (an overload, a server error, a dropped connection), llmwise sends the same request to Claude Haiku 4.5 through OpenRouter instead.
Anthropic's safety fallback	Anthropic's safety system can pass a Claude Opus 5.5 message to another Claude model. The reply then names the model that answered and says what you were charged for.	Doesn't apply

API prices are what our model catalog lists (Claude: Anthropic's list price). In llmwise you pay per message, not per token: the counts above are what you get.

Claude Opus 5.5 or Claude Haiku 4.5?

From the facts above and our test runs: the rest is how their answers suit your work, which one chat can show you.

Pick Claude Haiku 4.5: more messages for the money, up to 250 messages a month on Pro; you can try it in the free trial.

Each model's page, the families, and other pairs

Claude Opus 5.5 vs Claude Haiku 4.5 is one pair of models. The page below covers the whole family.

Both are made by the same company, so they're served the same way.

Questions

Is Claude Opus 5.5 or Claude Haiku 4.5 cheaper in llmwise?

Claude Haiku 4.5 gets four times as many messages: up to 250 messages a month on Pro, against up to 62 messages a month for Claude Opus 5.5. Every paid plan's monthly allowance is shared by all models, so each count is the most you get if it all goes to that model.

Can I try Claude Opus 5.5 and Claude Haiku 4.5 for free?

Claude Haiku 4.5 is, in the free trial of 5 messages; Claude Opus 5.5 needs a paid plan.

Which has the bigger context window, Claude Opus 5.5 or Claude Haiku 4.5?

Claude Opus 5.5: 1M tokens, against 200K tokens. A chat in llmwise holds up to 200k tokens, which fits in either, so the difference shows only through each maker's own API.

Can I use Claude Opus 5.5 and Claude Haiku 4.5 in the same chat?

Yes. Pick Claude Opus 5.5 for one message and Claude Haiku 4.5 for the next; the second sees the whole chat, including the first one's answer.

Which did better in your test runs, Claude Opus 5.5 or Claude Haiku 4.5?

On the same 50 prompts, run on September 27, 2026, Claude Opus 5.5 passed 49 and Claude Haiku 4.5 passed 43. The table on this page has each job, and every reply is published.

Claude, GPT, Gemini, DeepSeek, Grok, Kimi, and GLM, 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.