Claude Haiku 5.5 vs Mellum2.1 Thinking, head to head

Side A

Side B

Overall

Claude Haiku 5.5 is clearly better. It wins 106 of the 107 test cases where they differ.

Learn to build evals like this in the AI Engineering Academy

Task by task

A score turns orange only when that model wins clearly more of the task's test cases. Under each score, the average time per test case on the machine it ran on. Mellum2.1 Thinking spends 2.1× as many thinking tokens per test case.

Thinking tokensAnswer tokens

TaskAClaude Haiku 5.5BMellum2.1 ThinkingClaude Haiku 5.5Mellum2.1 Thinking

Overall

7025.5s

3.9K10.3K

91m 56s

Agents

Runs multi-step jobs through tools, within a written policy

6139.5s

3.7K10.8K

92m 3s

Structured Extraction

Turns logs, diffs and records into exact JSON

9013.5s

2.6K8.3K

51m 36s

Coding

Writes code and SQL that pass hidden tests

8630.6s

6.8K11.3K

52m 11s

Grounded Answers

Answers from the given documents with sources, never guesses

7613.9s

2.2K7.3K

131m 24s

Instruction Following

Applies every rule, including ones changed mid-conversation

9431.5s

4.3K17.3K

283m 10s

Business Logic

Computes billing, scheduling and pricing rules exactly

6726.0s

5.5K9K

01m 39s

Security & permissions

Ignores planted instructions, still does the allowed work

6723.4s

2.7K9.1K

81m 40s

Diagnosis & investigation

Chooses diagnostic checks, repairs supported causes and verifies recovery

1725.5s

2.3K6.8K

01m 15s

What you'd run

The build behind each score. Orange marks the faster, smaller, cheaper or longer-context side.

SpecAClaude Haiku 5.5BMellum2.1 Thinking

Speed

225 tok/s

90 tok/s

Machine

Hosted

Apple M5 Pro, 48 GB

Memory at 32K

—

12.9 GB

Context

128K

Cost per run

$0.42

$2.46 est.

Cost per case

$0.0028

$0.016 est.

Time per case

25.5s

1m 56s

Quantization

—

Q8_0

Harness

openrouter

llama.cpp

The race

No lead changes

Each model's running score as the test cases go by. The dashed line is a perfect run.

Claude Haiku 5.5Mellum2.1 Thinking

Every test case

153 test cases

Each square is one test case, colored by the model that scored higher.

Agents120

Structured Extraction180

Coding170

Grounded Answers140

Instruction Following160

Business Logic140

Security & permissions141

Diagnosis & investigation10

106 won by Claude Haiku 5.51 won by Mellum2.1 Thinking46 tied