GLM 5.3 Flash vs Underdog Saluki 27B, head to head

Side A

Side B

Overall

Too close to call. Each wins 30 of the 60 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 2.3× as many thinking tokens per test case.

Thinking tokensAnswer tokens

TaskAGLM 5.3 FlashBUnderdog Saluki 27BGLM 5.3 FlashUnderdog Saluki 27B

Overall

3555.5s

2.8K6.9K

335m 15s

Agents

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

1037.0s

60711.8K

409m 2s

Structured Extraction

Turns logs, diffs and records into exact JSON

4847.2s

2.7K4.4K

433m 17s

Coding

Writes code and SQL that pass hidden tests

541m 52s

5.3K9.2K

166m 54s

Grounded Answers

Answers from the given documents with sources, never guesses

3756.8s

2.9K4.5K

433m 22s

Instruction Following

Applies every rule, including ones changed mid-conversation

4340.8s

2K8K

466m 5s

Business Logic

Computes billing, scheduling and pricing rules exactly

331m 18s

5.8K5.2K

333m 55s

Security & permissions

Ignores planted instructions, still does the allowed work

5525.2s

9634.9K

433m 47s

Diagnosis & investigation

Chooses diagnostic checks, repairs supported causes and verifies recovery

027.0s

5638.7K

06m 39s

What you'd run

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

SpecAGLM 5.3 FlashBUnderdog Saluki 27B

Speed

60 tok/s

23 tok/s

Machine

Hosted

Apple M5 Pro, 48 GB

Memory at 32K

—

10.2 GB

Context

256K

Cost per run

$0.19

Not measured

Cost per case

$0.0013

Not measured

Time per case

55.5s

5m 15s

Quantization

FP8

IQ2_XS

Harness

openrouter

llama.cpp

The race

3 lead changes

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

GLM 5.3 FlashUnderdog Saluki 27B

Every test case

153 test cases

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

Agents17

Structured Extraction43

Coding122

Grounded Answers46

Instruction Following14

Business Logic44

Security & permissions44

Diagnosis & investigation00

30 won by GLM 5.3 Flash30 won by Underdog Saluki 27B93 tied