Beam 1.0 501B A23B vs Mellum2.1 Thinking, head to head

Side A

Side B

Overall

Beam 1.0 501B A23B is clearly better. It wins 87 of the 88 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 1.8× as many thinking tokens per test case.

Thinking tokensAnswer tokens

TaskABeam 1.0 501B A23BBMellum2.1 ThinkingBeam 1.0 501B A23BMellum2.1 Thinking

Overall

5717.2s

4K10.3K

91m 56s

Agents

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

5521.8s

4.8K10.8K

92m 3s

Structured Extraction

Turns logs, diffs and records into exact JSON

819.98s

3.4K8.3K

51m 36s

Coding

Writes code and SQL that pass hidden tests

4819.9s

4.5K11.3K

52m 11s

Grounded Answers

Answers from the given documents with sources, never guesses

6911.6s

2.7K7.3K

131m 24s

Instruction Following

Applies every rule, including ones changed mid-conversation

7820.9s

5.6K17.3K

283m 10s

Business Logic

Computes billing, scheduling and pricing rules exactly

5714.0s

4.2K9K

01m 39s

Security & permissions

Ignores planted instructions, still does the allowed work

5120.6s

3.2K9.1K

81m 40s

Diagnosis & investigation

Chooses diagnostic checks, repairs supported causes and verifies recovery

1724.3s

3.1K6.8K

01m 15s

What you'd run

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

SpecABeam 1.0 501B A23BBMellum2.1 Thinking

Speed

274 tok/s

90 tok/s

Machine

Hosted

Apple M5 Pro, 48 GB

Memory at 32K

—

12.9 GB

Context

128K

Cost per run

$0.83

$2.46 est.

Cost per case

$0.0054

$0.016 est.

Time per case

17.2s

1m 56s

Quantization

FP8

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.

Beam 1.0 501B A23BMellum2.1 Thinking

Every test case

153 test cases

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

Agents110

Structured Extraction160

Coding91

Grounded Answers130

Instruction Following130

Business Logic120

Security & permissions120

Diagnosis & investigation10

87 won by Beam 1.0 501B A23B1 won by Mellum2.1 Thinking65 tied