RankingEffort benchmarks

DeepSeek V4.1 Flash

Compare models

45 measured results · 6 effort settings · 25 benchmark/harness combinations

Change model

Compare with

Choose a second model and its effort to see both profiles and benchmark differences below.

Compare with

Capability profile

One configuration, seven capabilities

Comparison score · 0–100
DeepSeek V4.1 Flash · High capability radarComparison scores from 0 to 100. Unknown capabilities have no point; lines stop at gaps. Hollow points are preliminary. Exact values and support labels follow the chart.AgenticHard reasoningCodingHuman prefKnowledgeMultimodalLong context50100DeepSeek V4.1 Flash · High · Hard reasoning: 47.6 · SupportedDeepSeek V4.1 Flash · High · Coding: 83.9 · Preliminary

Gaps are unknown, not zero. Hollow points are preliminary.

  • Agentic

    Unknown
  • Hard reasoning

    A47.6Supported
  • Coding

    A83.9Preliminary
  • Human pref

    Unknown
  • Knowledge

    Unknown
  • Multimodal

    Unknown
  • Long context

    Unknown

Each profile belongs to the named configuration. Capability scores use a shared panel for all axes, distinct from the overall ranking panel; they are not accuracy percentages or direct head-to-head wins.

Capability score evidence

DeepSeek V4.1 Flash · High: 67.2 overall comparison score. Meets the overall evidence requirements.

2 areas measured. The available task mix affects this estimate; missing areas remain unknown.

Which sources carry the weight?
  • ARC Prize · 50.5% direct comparison weight · independent of the model provider
  • Cognition · 49.5% direct comparison weight · independent of the model provider
Which benchmark families carry the weight?
  • FrontierCode · 49.5% direct comparison weight
  • ARC-AGI · 30.2% direct comparison weight
  • arc-agi-3 · 20.3% direct comparison weight

These shares describe direct comparison weight, not fractions of the final score or confidence. Source balancing does not give every family equal weight or remove correlations between overlapping sources.

How sensitive is this score?

The audit removes one source at a time and refits against the same overall reference panel. These published results use the matching dataset audited 2026-11-22.

Both ranks use the same surviving configurations for each removal. The cohort starts with 602 configurations supported in the original audit, including separate effort settings. These are diagnostic positions, not the default leaderboard ranks. Eligibility is not rechecked after removal.

What changes when a source is removed?
Removed sourceRank before → afterRank movementCohort remaining / lost supportScore after (change)
Artificial Analysis60 → 59Up 1230 remaining / 372 lost67.5 (+0.3 points)
ARC Prize67 → 25Up 42602 remaining / 0 lost85.6 (+18.4 points)
Datacurve67 → 66Up 1602 remaining / 0 lost62.6 (-4.6 points)
LiveCodeBench67 → 67No movement602 remaining / 0 lost67.2 (-0.0 points)
Scale67 → 68Down 1602 remaining / 0 lost68.5 (+1.2 points)
Terminal-Bench67 → 68Down 1602 remaining / 0 lost67.6 (+0.3 points)
Cognition67 → 114Down 47602 remaining / 0 lost45.3 (-22.0 points)
LiveBench67 → 68Down 1601 remaining / 1 lost68.6 (+1.3 points)
SWE-rebench67 → 67No movement602 remaining / 0 lost66.8 (-0.5 points)
Epoch AI67 → 66Up 1588 remaining / 14 lost68.7 (+1.4 points)
Vals AI67 → 66Up 1601 remaining / 1 lost67.4 (+0.2 points)
Provider reports · Anthropic67 → 67No movement602 remaining / 0 lost67.2 (-0.0 points)
Provider reports · Moonshot AI67 → 67No movement602 remaining / 0 lost66.9 (-0.4 points)
Provider reports · Meta67 → 66Up 1602 remaining / 0 lost67.2 (-0.0 points)
Provider reports · Mistral67 → 67No movement602 remaining / 0 lost67.2 (+0.0 points)
Provider reports · OpenAI67 → 69Down 2602 remaining / 0 lost67.3 (+0.0 points)
All provider reports67 → 68Down 1602 remaining / 0 lost66.8 (-0.4 points)

Lost support means a configuration no longer has a comparison path to the complete reference panel. Its rank is unavailable. A surviving path alone does not establish leaderboard eligibility.

Overlapping sources may remain correlated even with source balancing. Leave-one-source-out movement is a sensitivity diagnostic, not a spread of source-specific ranks or a confidence measure. Scoring methodology · Dated configuration audit

Every benchmark family
Removed evidenceRecomputed scoreChange
frontiercode45.3-22.0
deepswe62.6-4.6
arc-agi71.0+3.8
arc-agi-369.8+2.5
mmmu-pro67.9+0.7
terminal-bench67.8+0.5
simpleqa67.7+0.5
swe-rebench66.8-0.5
hle67.7+0.4
enigma-eval67.7+0.4
mmlu-pro67.7+0.4
epoch-game-puzzles66.9-0.3
ifbench67.0-0.3
automationbench67.0-0.3
gpqa67.0-0.3
swe-bench-pro67.5+0.2
swe-atlas-qna67.0-0.2
multi-swe-bench67.0-0.2
gdp-pdf67.5+0.2
aa-lcr67.5+0.2
swe-atlas-test-writing67.1-0.2
vals-finance-agent67.1-0.2
tau-bench67.1-0.2
livebench-data67.4+0.2
aime67.1-0.2
vals-tax-agent67.1-0.1
livebench-instructions67.4+0.1
enterprise-ops67.4+0.1
vibe-code67.1-0.1
omniscience67.1-0.1
livebench-language67.3+0.1
vals-legal-research67.2-0.1
livebench-reasoning67.2-0.1
harvey67.3+0.1
livebench-coding67.2-0.1
apex-agents67.3+0.1
vals-excel-modeling67.2-0.1
terminal-bench-science67.2-0.1
vals-code-migration67.2-0.1
frontiermath67.2-0.0
gdpval67.3+0.0
swe-atlas-refactoring67.2-0.0
itbench67.2-0.0
livecodebench67.2-0.0
briefcase67.2-0.0
analyst-agent67.3+0.0
gmmlu67.2-0.0
math50067.2-0.0
critpt67.2-0.0
livebench-math67.2+0.0
What evidence is missing from the fit?

Across all collected settings for this model: 51 observations, 38 contributing, 0 matched without graph weight and 13 excluded. Counts do not establish rank eligibility.

  • 2 · Original SciCode grading defects are under review; retained as supporting evidence, excluded from aggregate and capability scores. Corrected SciCode-Verified requires a separate protocol review. Sources: https://arxiv.org/abs/2608.04975 and https://artificialanalysis.ai/evaluations/scicode
  • 1 · Benchmark family or source protocol has not been reviewed
  • 3 · Evaluator has not been reviewed for independent admission
  • 1 · Effort is not identified
  • 5 · Supporting evidence outside the reviewed capability core
  • 1 · No reviewed effort for this catalog observation; any separately collected effort measurement is counted separately
Inspect every observation and exclusion →

Benchmark results

These are published benchmark scores. Supported effort estimates also appear on the Capability leaderboard. Results from different harnesses occupy different rows. Missing cells stay unknown; no scores are copied between effort levels.

Reviewed retained observations are in use. October 2026 Real-SWE board has no 95% confidence whisker, so these scores stay at the 2026-10-18 retrieval. Original source retrieval 2026-10-18; source updated 2026-10-18.

Benchmark / sourceLowHighMaxNoneNon-reasoningUnspecified
AA-AnalystAgentArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Share of tasks solved on all 5 of 5 attempts · 80 tasks, 5 attempts per task · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-11-22.

Not reportedNot reported
35%
Reported settings

DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash

Not reportedNot reportedNot reported
AA-Briefcase EloArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

AA-Briefcase v1.1 is an agentic knowledge work benchmark developed by Artificial Analysis. AA-Briefcase Elo is a combined metric that aggregates rubric pass rate, analytical quality Elo and presentation Elo · Higher is better · Evaluation results measured independently by Artificial Analysis

Unit: elo. Source collected 2026-11-22.

Not reportedNot reported
1,415.04 Elo
Reported settings

DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash

Not reported
1,118.3 Elo
Reported settings

DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning

Not reported
AA-LCR v1.1Artificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-11-22.

Not reportedNot reported
84%
Reported settings

DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash

Not reported
55.33%
Reported settings

DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning

Not reported
AutomationBench-AAArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Share of task objectives completed with no guardrail violations · Higher is better · Benchmark developed by Zapier · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-11-22.

Not reportedNot reported
68.89%
Reported settings

DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash

Not reported
47.81%
Reported settings

DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning

Not reported
CritPtArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Benchmark developed by Argonne and UIUC, with contributions from 60+ researchers globally · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-11-22.

Not reportedNot reported
14.29%
Reported settings

DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash

Not reported
0.29%
Reported settings

DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning

Not reported
GDP.pdf: All-pass RateArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Share of attempts where every atomic criterion passed · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-11-22.

Not reportedNot reported
12.8%
Reported settings

DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash

Not reported
3.8%
Reported settings

DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning

Not reported
GDPval-AA v2.1 LeaderboardArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Elo rating for performance on real-world work tasks · Anchored to DeepSeek V4.1 Flash (max) at 1600 · Higher is better · Evaluation results measured independently by Artificial Analysis

Unit: elo. Source collected 2026-11-22.

Not reportedNot reported
1,600 Elo
Reported settings

DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash

Not reported
1,347.31 Elo
Reported settings

DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning

Not reported
Harvey LAB-AA v1.1: Hallucination-Gated All-Pass RateArtificial Analysis · Stirrup; Harvey private dataset v1.1.0; 120 tasks across 24 practice areas; up to 200 turns; internet disabled; GPT-6 Sol, Grok 4.7 and Claude Opus 5.5 rubric judges; GPT-6 Sol High two-pass hallucination check; hallucination-gated all-pass averaged across judges
Benchmark details

Average over tasks of the share of the judge panel finding every rubric criterion passed, with a task zeroed on any material hallucination · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-11-22.

Not reportedNot reported
1.67%
Reported settings

DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash

Not reportedNot reportedNot reported
Humanity's Last ExamArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-11-22.

Not reportedNot reported
39.25%
Reported settings

DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash

Not reported
10.8%
Reported settings

DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning

Not reported
ITBench-AAArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Combined public and private Site Reliability Engineering (SRE) tasks · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-11-22.

Not reportedNot reported
46.92%
Reported settings

DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash

Not reportedNot reportedNot reported
MLCR-AAArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Share of responses judged complete with no claim contradicting the record · Expert + compound (hard) sets · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-11-22.

Not reportedNot reported
23.89%
Reported settings

DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash

Not reportedNot reportedNot reported
MMMU-ProArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-11-22.

Not reportedNot reported
76.99%
Reported settings

DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash

Not reported
65.61%
Reported settings

DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning

Not reported
SciCodeArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings
Benchmark details

Benchmark curated by scientists across 16 disciplines. SciCode is under review following an independent audit of the dataset · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-11-22.

Not reportedNot reported
51.85%
Reported settings

DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash

Not reported
35.53%
Reported settings

DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning

Not reported
Terminal-Bench-Science 0.1Artificial Analysis · mini-swe-agent; Terminal-Bench-Science 0.1.0; 70 tasks; pass@1 averaged over 3 repeats
Benchmark details

Share of tasks passed (pass@1), averaged over three repeats · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-11-22.

Not reportedNot reported
9.05%
Reported settings

DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash

Not reported
0%
Reported settings

DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning

Not reported
Terminal-Bench v4.0Artificial Analysis · mini-SWE-agent v2.4.6; 66 tasks; pass@1 averaged over 3 repeats; 500 steps; task timeout up to 8 hours
Benchmark details

Benchmark developed by the Laude Institute, Stanford, and open-source contributors · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis

Unit: percent. Source collected 2026-11-22.

Not reportedNot reported
26.77%
Reported settings

DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash

Not reported
5.56%
Reported settings

DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning

Not reported
ARC-AGI-2ARC Prize · v2 Semi-Private
Benchmark details

Original public board. Configuration and task coverage may differ; inspect source.

Unit: percent. Source collected 2026-11-22.

60.56%
Reported settings

DeepSeek V4.1 Flash (Low); Source model ID: deepseek-v4-1-flash-low; 120 tasks

67.5%
Reported settings

DeepSeek V4.1 Flash (High); Source model ID: deepseek-v4-1-flash-high; 120 tasks

72.92%
Reported settings

DeepSeek V4.1 Flash (Max); Source model ID: deepseek-v4-1-flash-max; 120 tasks

1.25%
Reported settings

DeepSeek V4.1 Flash (None); Source model ID: deepseek-v4-1-flash-none; 120 tasks

Not reportedNot reported
ARC-AGI-3 · provider-adapterARC Prize · Provider Adapter
Benchmark details

ARC-AGI-3; Provider Adapter. Open each result for its exact effort, task subset, metric and protocol details.

Unit: percent. Source collected 2026-11-22.

3.26%
Reported settings

ARC-AGI-3 Semi-Private verified summary; Provider Adapter harness; source group deepseek-v4-1-flash; variant provider-adapter-low; public-demo task scores are excluded

3.15%
Reported settings

ARC-AGI-3 Semi-Private verified summary; Provider Adapter harness; source group deepseek-v4-1-flash; variant provider-adapter-high; public-demo task scores are excluded

2.68%
Reported settings

ARC-AGI-3 Semi-Private verified summary; Provider Adapter harness; source group deepseek-v4-1-flash; variant provider-adapter-max; public-demo task scores are excluded

0.48%
Reported settings

ARC-AGI-3 Semi-Private verified summary; Provider Adapter harness; source group deepseek-v4-1-flash; variant provider-adapter-none; public-demo task scores are excluded

Not reportedNot reported
ARC-AGI-3 · standardARC Prize · Standard
Benchmark details

ARC-AGI-3; Standard. Open each result for its exact effort, task subset, metric and protocol details.

Unit: percent. Source collected 2026-11-22.

0.35%
Reported settings

ARC-AGI-3 Semi-Private verified summary; Standard harness; source group deepseek-v4-1-flash; variant low; public-demo task scores are excluded

0.48%
Reported settings

ARC-AGI-3 Semi-Private verified summary; Standard harness; source group deepseek-v4-1-flash; variant high; public-demo task scores are excluded

0.85%
Reported settings

ARC-AGI-3 Semi-Private verified summary; Standard harness; source group deepseek-v4-1-flash; variant max; public-demo task scores are excluded

0.34%
Reported settings

ARC-AGI-3 Semi-Private verified summary; Standard harness; source group deepseek-v4-1-flash; variant none; public-demo task scores are excluded

Not reportedNot reported
FrontierCode 1.1 · extended-devin-cliCognition · devin-cli
Benchmark details

FrontierCode 1.1; devin-cli. Open each result for its exact effort, task subset, metric and protocol details.

Unit: percent. Source collected 2026-11-22.

Not reported
56.62%
Reported settings

FrontierCode 1.1; extended; 150 tasks; devin-cli; source effort high; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0.0049

Not reportedNot reportedNot reportedNot reported
FrontierCode 1.1 · main-devin-cliCognition · devin-cli
Benchmark details

FrontierCode 1.1; devin-cli. Open each result for its exact effort, task subset, metric and protocol details.

Unit: percent. Source collected 2026-11-22.

Not reported
43.87%
Reported settings

FrontierCode 1.1; main; 100 tasks; devin-cli; source effort high; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0.0073

Not reportedNot reportedNot reportedNot reported
Artificial Analysis Cyber IndexArtificial Analysis · Artificial Analysis Stirrup; defensive code audit, patch and end-to-end vulnerability tasks
Benchmark details

v1.0. Equal mean of CWE-Bench-AA, DeepsecBench-AA and CyberGym-E2E-AA. Display only. This composite measures defensive source-code tasks. Successful capability scores depend on safety refusals, and refusal rates are reported separately. It does not enter weighted Overall. https://artificialanalysis.ai/methodology/cyber-index

Unit: index. Source collected 2026-11-22.

Not reportedNot reported
40.81
Reported settings

Artificial Analysis Stirrup; v1.0; DeepSeek V4.1 Flash (Max); source label DeepSeek V4.1 Flash (Max); mitigation standard; public model; source field cyberIndex; index 0–100; CWE-Bench-AA pass@1 0.683333333333333; refusal 0; DeepsecBench-AA median F2, 3 repeats, GPT-5.6 Sol High judge 0.31199859972578; refusal 0; CyberGym-E2E-AA pass@1 0.22900763358778625; refusal 0; https://artificialanalysis.ai/models/deepseek-v4-1-flash; page sha256 344ca66f5aa3f3abd1901b12bd29db73950d90e553fcb3ae900dcf590d7a6db2; encrypted payload sha256 b5e6ae259181eaed471e288c4bbdbd1088438e0a81caf19ec0fa60f7092d48f7; Display only. This composite measures defensive source-code tasks. Successful capability scores depend on safety refusals, and refusal rates are reported separately. It does not enter weighted Overall.

Not reportedNot reportedNot reported
LMArena Agent ArenaLMArena · Arena Agent Mode. The board does not name a per-model scaffold.
Benchmark details

Aggregate Agent Arena IPS net improvement, stored as the board's printed percentage points. Arena Agent Mode does not name a per-model scaffold. Display only. Excluded from weighted Overall.

Unit: index. Source collected 2026-11-22.

Not reportedNot reported
4.02
Reported settings

Deepseek V4.1 Flash (Max); Arena Agent Mode; IPS net improvement 4.02 points ±0.47. rank 17; observations 16932092; sessions 112938; published 2026-10-02. The board does not name a per-model scaffold. Display only. Excluded from weighted Overall.

Not reportedNot reportedNot reported
FrontierCS 2FrontierCS · Vendor harness named on each row. The official data file does not publish an effort setting.
Benchmark details

Pass rate on 48 tasks. 1 run per task. Each row uses that model's vendor harness. Kaggle verification is unfinished. UC Berkeley and Princeton. Display only. Excluded from weighted Overall. https://arxiv.org/abs/2610.29932 https://www.kaggle.com/benchmarks/frontiercs/frontiercs-2

Unit: percent. Source collected 2026-11-22.

Not reportedNot reportedNot reportedNot reportedNot reported
14.6%
Reported settings

DeepSeek V4.1 Flash; DeepSeek Harness; effort unspecified; 7 of 48 tasks passed; 1 run per task. Each row uses that model's vendor harness. Kaggle verification is unfinished. UC Berkeley and Princeton. Display only. Excluded from weighted Overall.

MaxxBenchunLAP · unLAP public harness, Pier 0.3.1, 89 mutated DeepSWE tasks.
Benchmark details

unLAP MaxxBench. 89 mutated DeepSWE tasks, 2 completed evaluations, 178 attempts. The score is the published percent of passed attempts. Harness schema 2.0 at https://github.com/unlap-labs/maxxbench/commit/1a0df5e9c5d4bccbae131b01830d478088fea947. The board publishes no numeric version. Display only. Excluded from weighted Overall. Not a DeepSWE score.

Unit: percent. Source collected 2026-11-22.

Not reportedNot reported
55.6%
Reported settings

DeepSeek V4.1 Flash max; unLAP MaxxBench. 89 mutated DeepSWE tasks, 2 completed evaluations, 178 attempts. The score is the published percent of passed attempts. Harness schema 2.0 at https://github.com/unlap-labs/maxxbench/commit/1a0df5e9c5d4bccbae131b01830d478088fea947. The board publishes no numeric version. Display only. Excluded from weighted Overall. Not a DeepSWE score. Passed 99 of 178. Capture date is not an evaluation date.

Not reportedNot reportedNot reported
Multi-step SWEKlavis AI · Harbor harness named on each row, at the highest effort that harness supports.
Benchmark details

Solve rate on 69 chained tasks and 574 steps. Solve rate counts a rollout only when every step receives full credit. 95% bootstrap interval over tasks. Harbor multi-step format. Klavis AI v1.0. Display only. Excluded from weighted Overall. https://arxiv.org/abs/2611.07166 https://github.com/Klavis-AI/Multi-step-SWE

Unit: percent. Source collected 2026-11-22.

Not reportedNot reported
6.5%
Reported settings

DeepSeek V4.1 Flash; Claude Code; effort max; solve rate 6.5% [2.9, 10.9]; 18 of 276 rollouts solved; steps passed 35.8%; Pass@4 15.9; 69 tasks; 574 steps; Solve rate counts a rollout only when every step receives full credit. 95% bootstrap interval over tasks. Harbor multi-step format. Klavis AI v1.0. Display only. Excluded from weighted Overall.

Not reportedNot reportedNot reported
What these comparisons mean

Each source reports its own effort labels. Matching labels do not establish equal compute budgets or identical fallback behavior. Open Reported settings to inspect the exact model name, agent and task coverage. Multiple results at one level remain visible rather than selecting the best. Source-reported None or Non-reasoning labels are not treated as evidence that an API supports those settings.

These measurements use their own dated source collection. The leaderboard fits identified configurations jointly; it contains no pooled model ratings. Explicit thinking budgets are separate configurations. Read the methodology.

Download all effort benchmark results · 10,184 results across 381 models · Collected 2026-11-22