Beam 1.0 501B A23B vs Clover 35B A3B Preview, head to head

Side A

Side B

Overall

Beam 1.0 501B A23B is clearly better. It wins 57 of the 69 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. The time under each score is the average per test case, on the machine it ran on. Clover 35B A3B Preview spends 2.3× as many thinking tokens per test case.

Thinking tokensAnswer tokens

TaskABeam 1.0 501B A23BBClover 35B A3B PreviewBeam 1.0 501B A23BClover 35B A3B Preview

Overall

5717.2s

4K8.4K

3035.1s

Agents

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

5521.8s

4.8K9.6K

2036.8s

Structured Extraction

Turns logs, diffs and records into exact JSON

819.98s

3.4K7.2K

4327.5s

Coding

Writes code and SQL that pass hidden tests

4819.9s

4.5K12.2K

101m 5s

Grounded Answers

Answers from the given documents with sources, never guesses

6911.6s

2.7K5.4K

4821.0s

Instruction Following

Applies every rule, including ones changed mid-conversation

7820.9s

5.6K9.2K

5332.8s

Business Logic

Computes billing, scheduling and pricing rules exactly

5714.0s

4.2K9.8K

1936.1s

Security & permissions

Ignores planted instructions, still does the allowed work

5120.6s

3.2K5.8K

4725.6s

Diagnosis & investigation

Chooses diagnostic checks, repairs supported causes and verifies recovery

1724.3s

3.1K7K

037.0s

What you'd run

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

SpecABeam 1.0 501B A23BBClover 35B A3B Preview

Speed

274 tok/s

252 tok/s

Machine

Hosted

Memory at 32K

—

Context

—

Cost per run

$0.83

Not measured

Cost per case

$0.0054

Not measured

Time per case

17.2s

35.1s

Quantization

FP8

—

Harness

openrouter

custom

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 A23BClover 35B A3B Preview

Every test case

153 test cases

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

Agents112

Structured Extraction102

Coding80

Grounded Answers61

Instruction Following91

Business Logic91

Security & permissions35

Diagnosis & investigation10

57 won by Beam 1.0 501B A23B12 won by Clover 35B A3B Preview84 tied