# DeepSeek V4.1 Flash max vs Pareto 26.9 medium for coding

Stet publishes a public search API over its indexed pages; the guide is at https://www.stet.sh/agents/agent-access.

Source: https://www.stet.sh/leaderboard/compare/deepseek-v4-1-flash-max-vs-pareto-26-9-medium
Source observed: 2026-11-19T11:08:21.081Z
Last checked: 2026-11-19T11:08:21.081+00:00
Indexed: 2026-11-19T11:19:06.375+00:00

---

# DeepSeek V4.1 Flash max vs Pareto 26.9 medium

我们比较两个不同 harness 的编码 agent 配置：DeepSeek V4.1 Flash max 在 `deepseek-harness`、max reasoning 中运行；Pareto 26.9 medium 在 `claude-code`、medium reasoning 中运行。结果比较的是 setups，不是模型单独的能力。

## Headline result

- **Quality:** 没有明确差异。DeepSeek 赢 4 个 head-to-head，Pareto 赢 3 个，11 个打平，6 个未决。
- **Speed:** Pareto **1.3× faster**：39.2 versus 49.3 分钟 per task。
- **Cost:** DeepSeek **12× cheaper**：$0.49 versus $5.98 per task；DeepSeek 的 $0.49 来自 23/24 tasks。

## What this comparison is

- One private real repository: `Stet-AI/stet`; not public benchmark scores.
- 24 merged changes, replayed end to end from the repository state before each change.
- The repository’s tests and judges comparing each patch with the merged change graded the work.
- We measured quality, cost, and time on real agent work; the sample is smaller than public benchmarks.

Data: November 19, 2026. Both setups came from leaderboard run `bmk_d4ba310fdf50db4ccb2092e3` of 18 setups.

## Stet score and rank

| Setup | Harness | Reasoning | Rank | Stet score | 95% range |
| --- | --- | --- | --- | --- | --- |
| DeepSeek V4.1 Flash max | `deepseek-harness` | max | 8 of 18 | 51.0% | 42.1–60.1 |
| Pareto 26.9 medium | `claude-code` | medium | 7 of 18 | 52.3% | 44.9–59.5 |

The 95% score ranges overlap. DeepSeek’s win share was 52%, with a 95% range of 37–66%; ties are within judge noise.

## Other quality measures

These measures are not part of the headline Stet score and have no confidence interval.

| Metric | DeepSeek V4.1 Flash max | Pareto 26.9 medium |
| --- | --- | --- |
| Code review pass rate | 33% — 24 tasks | 35% — 23 tasks |
| Checklist score | 0.83 — 24 tasks | 0.80 — 24 tasks |
| Equivalence rate | 43% — 23 tasks | 39% — 23 tasks |

The higher recorded value is not a proven lead.

## Task-by-task outcomes

DeepSeek solved 3 tasks that Pareto did not; Pareto solved 2 that DeepSeek did not. Both solved 7; neither solved 6. On the 7 both solved, judges preferred DeepSeek’s patch on 1, Pareto’s on 1, and called 5 even. Six of the 24 tasks are left out because one setup has no scored result.

### Only DeepSeek V4.1 Flash max solved

- `stet-15439c21` — 5 changed lines in 2 files; judges preferred DeepSeek’s patch.
- `stet-commit-5d766dde` — 218 changed lines in 2 files; judges preferred DeepSeek’s patch.
- `stet-commit-342d0ac0` — 640 changed lines in 8 files; judges preferred DeepSeek’s patch.

### Only Pareto 26.9 medium solved

- `stet-2450ca2d` — 2 changed lines in 1 file; judges preferred Pareto’s patch.
- `stet-commit-f0caded9` — 113 changed lines in 4 files; judges preferred Pareto’s patch.

### Both solved

- `stet-43493c52` — 4 changed lines in 1 file; judges preferred Pareto’s patch.
- `stet-89dfbc27` — 9 changed lines in 1 file.
- `stet-232b50ba` — 10 changed lines in 1 file.
- `stet-5276aa3f` — 23 changed lines in 2 files.
- `stet-b28f4fe7` — 29 changed lines in 1 file.
- `stet-66762b4c` — 84 changed lines in 1 file; judges preferred DeepSeek’s patch.
- `stet-commit-f8ae1177` — 1,312 changed lines in 9 files.

### Neither solved

- `stet-6f84e978` — 23 changed lines in 1 file.
- `stet-commit-e928166f` — 149 changed lines in 2 files.
- `stet-commit-30506f8e` — 476 changed lines in 17 files.
- `stet-commit-60a0e6c2` — 552 changed lines in 11 files.
- `stet-commit-87814583` — 763 changed lines in 8 files.
- `stet-commit-2826765e` — 1,420 changed lines in 19 files.

A task is solved when tests passed or a judge found the fix equivalent to the merged change. Stet score also weighs code quality.

## All 18 compared tasks

| Task | DeepSeek V4.1 Flash max | Pareto 26.9 medium | Solved by | Better patch |
| --- | --- | --- | --- | --- |
| `stet-89dfbc27` · 9 lines, 1 file | Passed; $0.09 · 9.3 min | Passed; $3.94 · 47.2 min | Both | Even |
| `stet-b28f4fe7` · 29 lines, 1 file | Passed; $0.22 · 21.5 min | Passed; $3.17 · 22.7 min | Both | Even |
| `stet-232b50ba` · 10 lines, 1 file | Passed; $0.34 · 59.2 min | Passed; $3.09 · 42.8 min | Both | Even |
| `stet-5276aa3f` · 23 lines, 2 files | Passed; $0.16 · 37.9 min | Passed; $0.69 · 13.0 min | Both | Even |
| `stet-66762b4c` · 84 lines, 1 file | Passed; $0.09 · 15.1 min | Passed; $3.39 · 50.6 min | Both | DeepSeek |
| `stet-commit-5d766dde` · 218 lines, 2 files | Equivalent; $0.27 · 35.4 min | Failed; $1.02 · 7.0 min | DeepSeek only | DeepSeek |
| `stet-43493c52` · 4 lines, 1 file | Passed; $0.22 · 63.4 min | Passed; $2.04 · 29.5 min | Both | Pareto |
| `stet-commit-f0caded9` · 113 lines, 4 files | Failed; $0.25 · 42.5 min | Equivalent; $8.97 · 60.8 min | Pareto only | Pareto |
| `stet-15439c21` · 5 lines, 2 files | Passed; $0.15 · 45.6 min | Failed; $2.73 · 36.0 min | DeepSeek only | DeepSeek |
| `stet-commit-f8ae1177` · 1,312 lines, 9 files | Passed; $1.72 · 113.6 min | Passed; $9.84 · 52.6 min | Both | Even |
| `stet-2450ca2d` · 2 lines, 1 file | Failed; $0.15 · 29.5 min | Passed; $0.61 · 12.3 min | Pareto only | Pareto |
| `stet-commit-342d0ac0` · 640 lines, 8 files | Equivalent; $1.05 · 59.4 min | Failed; $7.95 · 42.6 min | DeepSeek only | DeepSeek |
| `stet-6f84e978` · 23 lines, 1 file | Failed; $0.24 · 28.7 min | Failed; $2.15 · 20.9 min | Neither | Even |
| `stet-commit-60a0e6c2` · 552 lines, 11 files | Failed; $0.58 · 47.5 min | Failed; $5.64 · 20.5 min | Neither | Even |
| `stet-commit-2826765e` · 1,420 lines, 19 files | Failed; $0.80 · 70.4 min | Failed; $8.80 · 40.3 min | Neither | Even |
| `stet-commit-e928166f` · 149 lines, 2 files | Failed; $0.58 · 58.7 min | Failed; $7.09 · 34.0 min | Neither | Even |
| `stet-commit-30506f8e` · 476 lines, 17 files | Failed; 71.2 min | Failed; $16.87 · 66.0 min | Neither | Even |
| `stet-commit-87814583` · 763 lines, 8 files | Failed; $0.38 · 61.5 min | Failed; $11.99 · 89.0 min | Neither | Even |

## Cost and time

- **DeepSeek cost:** mean $0.43, median $0.25, from 17/18 tasks; side-by-side cost is $0.49 from 23/24 tasks.
- **Pareto cost:** mean $5.56, median $3.66.
- DeepSeek cost less on all 17 tasks with recorded paired cost.
- **DeepSeek agent time:** mean 48.4 minutes, median 46.6 minutes.
- **Pareto agent time:** mean 38.2 minutes, median 38.2 minutes.
- Pareto finished faster on 13 of 18 tasks; DeepSeek on 5.

## Where each setup may fit

### Small fixes

10 tasks; 2–84 changed lines in 1–2 files. Both solved 7/9. DeepSeek cost 13× less per task; Pareto was 1.1× faster.

| Metric | DeepSeek V4.1 Flash max | Pareto 26.9 medium |
| --- | ---: | ---: |
| Solved | 7/9 | 7/9 |
| Cost per task | $0.18 | $2.42 |
| Agent time per task | 34.5 min | 30.6 min |

### Large, cross-cutting changes

14 tasks; 113–1,420 changed lines in up to 19 files. DeepSeek solved 3/9 versus Pareto’s 2/9; Pareto was 1.4× faster.

| Metric | DeepSeek V4.1 Flash max | Pareto 26.9 medium |
| --- | ---: | ---: |
| Solved | 3/9 | 2/9 |
| Cost per task | $0.70 from 8/9 tasks | $8.69 |
| Agent time per task | 62.3 min | 45.9 min |

## Side by side

| Metric | DeepSeek V4.1 Flash max | Pareto 26.9 medium |
| --- | --- | --- |
| Harness | `deepseek-harness` | `claude-code` |
| Reasoning effort | max | medium |
| Rank | 8 of 18 | 7 of 18 |
| Stet score | 51.0%; 95% range 42.1–60.1 | 52.3%; 95% range 44.9–59.5 |
| Solved | 10/24 — 8 passed, 2 judged equivalent | 11/24 — 10 passed, 1 judged equivalent |
| Cost per task | $0.49 from 23/24 tasks | $5.98 |
| Agent time per task | 49.3 min | 39.2 min |

## Research links

- [Cost per successful task ranked the coding agent that lost every head to head 1st](https://www.stet.sh/blog/cost-per-successful-task)
- [A benchmark mined from my own repo agreed with FrontierCode](https://www.stet.sh/blog/benchmark-agreed-with-frontier-code)
- [See all 18 setups](https://www.stet.sh/leaderboard)
- [How Stet measures](https://www.stet.sh/methodology)
- [Get started](https://www.stet.sh/private)

## FAQ

**Which setup is better?** Quality was not clearly different: DeepSeek won 4, Pareto won 3, 11 tied, and 6 were undecided. DeepSeek was 12× cheaper; Pareto was 1.3× faster.

**How did we test them?** Both agents worked on the same 24 merged changes from private `Stet-AI/stet`, starting from the repository state before each change. Repository tests and judges graded each patch against the merged change; no public benchmark scores were used.

We replay your merged changes, score quality above pass/fail, and deliver comparison reports through your agent. [Get started](https://www.stet.sh/private)

---

[Read this page’s summary](https://www.stet.sh/agents/api/public/v1/page-tldr?url=https%3A%2F%2Fwww.stet.sh%2Fleaderboard%2Fcompare%2Fdeepseek-v4-1-flash-max-vs-pareto-26-9-medium&trace=t_5da45423e484)

## Ask a follow-up about this page

Need a more specific answer? Ask the exact question you have. It returns a fresh, cited answer from Stet’s knowledge base. No account or API key required.

Open this URL with your own question, URL-encoded. Include your question in q. All research inputs are optional; no account or setup.

`GET https://www.stet.sh/agents/agent/faq?q={question}&trace=t_5da45423e484`

[Tool guide and examples](https://www.stet.sh/agents/agent-access) · [OpenAPI](https://www.stet.sh/agents/api/public/v1/openapi.json) · [Page TLDR form](https://www.stet.sh/agents/agent-tools/page-tldr?url=https%3A%2F%2Fwww.stet.sh%2Fleaderboard%2Fcompare%2Fdeepseek-v4-1-flash-max-vs-pareto-26-9-medium). Opening a form does not execute a tool.