Flint vs Qwen3.8 27B, head to head

Side A

Side B

Overall

Qwen3.8 27B is clearly better. It wins 60 of the 72 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. Flint spends 1.8× as many thinking tokens per test case.

Thinking tokensAnswer tokens

TaskAFlintBQwen3.8 27BFlintQwen3.8 27B

Overall

211m 17s

10.9K5.9K

4833.3s

Agents

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

1447.7s

4.8K5.2K

5736.6s

Structured Extraction

Turns logs, diffs and records into exact JSON

291m 17s

10.9K4.6K

7320.6s

Coding

Writes code and SQL that pass hidden tests

141m 19s

12K9.3K

3753.0s

Grounded Answers

Answers from the given documents with sources, never guesses

2957.7s

8.7K4.2K

6131.6s

Instruction Following

Applies every rule, including ones changed mid-conversation

352m 28s

22.2K6.9K

6733.6s

Business Logic

Computes billing, scheduling and pricing rules exactly

101m 20s

11.7K7.6K

3632.9s

Security & permissions

Ignores planted instructions, still does the allowed work

3552.6s

7.1K3.7K

5122.6s

Diagnosis & investigation

Chooses diagnostic checks, repairs supported causes and verifies recovery

057.0s

7.3K5.9K

040.0s

What you'd run

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

SpecAFlintBQwen3.8 27B

Speed

149 tok/s

196 tok/s

Machine

Hosted

Memory at 32K

—

Context

256K

Cost per run

Not measured

$1.28

Cost per case

Not measured

$0.0083

Time per case

1m 17s

33.3s

Quantization

BF16

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.

FlintQwen3.8 27B

Every test case

153 test cases

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

Agents010

Structured Extraction312

Coding07

Grounded Answers18

Instruction Following59

Business Logic18

Security & permissions26

Diagnosis & investigation00

12 won by Flint60 won by Qwen3.8 27B81 tied