GLM 5.3 Flash vs Qwen3.8 27B, head to head

Side A

Side B

Overall

Qwen3.8 27B is clearly better. It wins 50 of the 81 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. Qwen3.8 27B spends 2.2× as many thinking tokens per test case.

Thinking tokensAnswer tokens

TaskAGLM 5.3 FlashBQwen3.8 27BGLM 5.3 FlashQwen3.8 27B

Overall

3555.5s

2.8K5.9K

4833.3s

Agents

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

1037.0s

6075.2K

5736.6s

Structured Extraction

Turns logs, diffs and records into exact JSON

4847.2s

2.7K4.6K

7320.6s

Coding

Writes code and SQL that pass hidden tests

541m 52s

5.3K9.3K

3753.0s

Grounded Answers

Answers from the given documents with sources, never guesses

3756.8s

2.9K4.2K

6131.6s

Instruction Following

Applies every rule, including ones changed mid-conversation

4340.8s

2K6.9K

6733.6s

Business Logic

Computes billing, scheduling and pricing rules exactly

331m 18s

5.8K7.6K

3632.9s

Security & permissions

Ignores planted instructions, still does the allowed work

5525.2s

9633.7K

5122.6s

Diagnosis & investigation

Chooses diagnostic checks, repairs supported causes and verifies recovery

027.0s

5635.9K

040.0s

What you'd run

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

SpecAGLM 5.3 FlashBQwen3.8 27B

Speed

60 tok/s

196 tok/s

Machine

Hosted

Memory at 32K

—

Context

256K

Cost per run

$0.19

$1.28

Cost per case

$0.0013

$0.0083

Time per case

55.5s

33.3s

Quantization

FP8

Harness

openrouter

The race

No lead changes

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

GLM 5.3 FlashQwen3.8 27B

Every test case

153 test cases

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

Agents010

Structured Extraction511

Coding73

Grounded Answers48

Instruction Following49

Business Logic65

Security & permissions54

Diagnosis & investigation00

31 won by GLM 5.3 Flash50 won by Qwen3.8 27B72 tied