Muse Spark 1.3 Contributor vs Underdog Saluki 27B, head to head

Side A

Side B

Overall

Muse Spark 1.3 Contributor is clearly better. It wins 61 of the 67 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. Muse Spark 1.3 Contributor spends 1.4× as many thinking tokens per test case.

Thinking tokensAnswer tokens

TaskAMuse Spark 1.3 ContributorBUnderdog Saluki 27BMuse Spark 1.3 ContributorUnderdog Saluki 27B

Overall

661m 42s

8.7K6.9K

335m 15s

Agents

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

573m 16s

14.7K11.8K

409m 2s

Structured Extraction

Turns logs, diffs and records into exact JSON

9037.9s

4.2K4.4K

433m 17s

Coding

Writes code and SQL that pass hidden tests

482m 1s

10.8K9.2K

166m 54s

Grounded Answers

Answers from the given documents with sources, never guesses

811m 1s

6.1K4.5K

433m 22s

Instruction Following

Applies every rule, including ones changed mid-conversation

841m 18s

8.7K8K

466m 5s

Business Logic

Computes billing, scheduling and pricing rules exactly

761m 7s

6K5.2K

333m 55s

Security & permissions

Ignores planted instructions, still does the allowed work

761m 59s

8.2K4.9K

433m 47s

Diagnosis & investigation

Chooses diagnostic checks, repairs supported causes and verifies recovery

173m 34s

17.2K8.7K

06m 39s

What you'd run

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

SpecAMuse Spark 1.3 ContributorBUnderdog Saluki 27B

Speed

107 tok/s

23 tok/s

Machine

Hosted

Apple M5 Pro, 48 GB

Memory at 32K

—

10.2 GB

Context

256K

Cost per run

$0.41

Not measured

Cost per case

$0.0027

Not measured

Time per case

1m 42s

5m 15s

Quantization

—

IQ2_XS

Harness

openrouter

llama.cpp

The race

1 lead change

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

Muse Spark 1.3 ContributorUnderdog Saluki 27B

Every test case

153 test cases

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

Agents82

Structured Extraction100

Coding93

Grounded Answers91

Instruction Following80

Business Logic90

Security & permissions70

Diagnosis & investigation10

61 won by Muse Spark 1.3 Contributor6 won by Underdog Saluki 27B86 tied