Mistral Large 4 vs Underdog Saluki 27B, head to head

Side A

Side B

Overall

Underdog Saluki 27B is clearly better. It wins 28 of the 44 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. Underdog Saluki 27B spends 1.3× as many thinking tokens per test case.

Thinking tokensAnswer tokens

TaskAMistral Large 4BUnderdog Saluki 27BMistral Large 4Underdog Saluki 27B

Overall

281m 56s

6.2K6.9K

335m 15s

Agents

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

1126.1s

79511.8K

409m 2s

Structured Extraction

Turns logs, diffs and records into exact JSON

153.78s

2884.4K

433m 17s

Coding

Writes code and SQL that pass hidden tests

332m 48s

9.3K9.2K

166m 54s

Grounded Answers

Answers from the given documents with sources, never guesses

382m 45s

8.1K4.5K

433m 22s

Instruction Following

Applies every rule, including ones changed mid-conversation

693m 47s

12.9K8K

466m 5s

Business Logic

Computes billing, scheduling and pricing rules exactly

242m 40s

8.8K5.2K

333m 55s

Security & permissions

Ignores planted instructions, still does the allowed work

331m 9s

3.7K4.9K

433m 47s

Diagnosis & investigation

Chooses diagnostic checks, repairs supported causes and verifies recovery

01m 19s

4.4K8.7K

06m 39s

What you'd run

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

SpecAMistral Large 4BUnderdog Saluki 27B

Speed

56 tok/s

23 tok/s

Machine

Hosted

Apple M5 Pro, 48 GB

Memory at 32K

—

10.2 GB

Context

512K

256K

Cost per run

$2.67

Not measured

Cost per case

$0.017

Not measured

Time per case

1m 56s

5m 15s

Quantization

—

IQ2_XS

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.

Mistral Large 4Underdog Saluki 27B

Every test case

153 test cases

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

Agents17

Structured Extraction19

Coding74

Grounded Answers23

Instruction Following51

Business Logic02

Security & permissions02

Diagnosis & investigation00

16 won by Mistral Large 428 won by Underdog Saluki 27B109 tied