DeepSeek V4.1 Flash 1017 vs Flint, head to head

Side A

Side B

Overall

DeepSeek V4.1 Flash 1017 is clearly better. It wins 68 of the 75 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 2.0× as many thinking tokens per test case.

Thinking tokensAnswer tokens

TaskADeepSeek V4.1 Flash 1017BFlintDeepSeek V4.1 Flash 1017Flint

Overall

5626.0s

5.3K10.9K

211m 17s

Agents

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

5826.7s

3.3K4.8K

1447.7s

Structured Extraction

Turns logs, diffs and records into exact JSON

8117.2s

4.2K10.9K

291m 17s

Coding

Writes code and SQL that pass hidden tests

4840.8s

9.4K12K

141m 19s

Grounded Answers

Answers from the given documents with sources, never guesses

7119.8s

4.4K8.7K

2957.7s

Instruction Following

Applies every rule, including ones changed mid-conversation

7625.6s

5.5K22.2K

352m 28s

Business Logic

Computes billing, scheduling and pricing rules exactly

6726.6s

6.6K11.7K

101m 20s

Security & permissions

Ignores planted instructions, still does the allowed work

4826.6s

4.7K7.1K

3552.6s

Diagnosis & investigation

Chooses diagnostic checks, repairs supported causes and verifies recovery

020.2s

3.2K7.3K

057.0s

What you'd run

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

SpecADeepSeek V4.1 Flash 1017BFlint

Speed

243 tok/s

149 tok/s

Machine

Hosted

Memory at 32K

—

Context

256K

Cost per run

$0.75

Not measured

Cost per case

$0.0049

Not measured

Time per case

26.0s

1m 17s

Quantization

—

BF16

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.

DeepSeek V4.1 Flash 1017Flint

Every test case

153 test cases

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

Agents110

Structured Extraction121

Coding70

Grounded Answers90

Instruction Following102

Business Logic131

Security & permissions63

Diagnosis & investigation00

68 won by DeepSeek V4.1 Flash 10177 won by Flint78 tied