RankingEffort benchmarks

DeepSeek V4.1 Flash

Compare models

27 measured results · 2 effort settings · 15 benchmark/harness combinations

Model evidence and sources →

Change model

Model

Current configurationSources

Compare with

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

Compare withModel and configuration

Capability profile

One configuration, seven capabilities

Comparison score · 0–100

ADeepSeek V4.1 Flash · Max2 of 7 capabilities supported

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

Agentic
A60.2Supported
Hard reasoning
A41.8Supported
Coding
A67.2Preliminary
Human pref
Unknown
Knowledge
A41.4Preliminary
Multimodal
A30.0Preliminary
Long context
A68.2Preliminary

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 · Max: 48.1 overall comparison score. Meets the overall evidence requirements.

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

Which sources carry the weight?

Artificial Analysis · 100.0% direct comparison weight · independent of the model provider

Which benchmark families carry the weight?

Aa Lcr · 11.6% direct comparison weight
Humanity’s Last Exam · 10.3% direct comparison weight
Critpt · 10.2% direct comparison weight
omniscience · 10.1% direct comparison weight
Mmmu Pro · 10.0% direct comparison weight
Other families (7) · 47.8% 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 2026-10-02.

Both ranks use the same surviving configurations for each removal. The cohort starts with 434 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 source	Rank before → after	Rank movement	Cohort remaining / lost support	Score after (change)
Artificial Analysis	Comparison support lost; no rank		197 remaining / 237 lost	Unavailable
ARC Prize	63 → 61	Up 2	434 remaining / 0 lost	53.6 (+5.5 points)
Datacurve	63 → 65	Down 2	434 remaining / 0 lost	43.9 (-4.2 points)
LiveCodeBench	63 → 63	No movement	434 remaining / 0 lost	48.0 (-0.1 points)
Scale	63 → 64	Down 1	434 remaining / 0 lost	50.0 (+1.9 points)
Terminal-Bench	63 → 63	No movement	434 remaining / 0 lost	48.3 (+0.2 points)
Cognition	63 → 66	Down 3	434 remaining / 0 lost	47.6 (-0.5 points)
LiveBench	63 → 63	No movement	433 remaining / 1 lost	49.1 (+1.0 points)
SWE-rebench	63 → 63	No movement	434 remaining / 0 lost	48.2 (+0.1 points)
Epoch AI	62 → 64	Down 2	416 remaining / 18 lost	48.2 (+0.1 points)
Vals AI	63 → 64	Down 1	433 remaining / 1 lost	47.1 (-1.0 points)
Provider reports · Anthropic	63 → 63	No movement	434 remaining / 0 lost	48.1 (-0.0 points)
Provider reports · Moonshot AI	63 → 63	No movement	434 remaining / 0 lost	47.7 (-0.4 points)
Provider reports · Meta	63 → 64	Down 1	434 remaining / 0 lost	48.0 (-0.1 points)
Provider reports · Mistral	63 → 63	No movement	434 remaining / 0 lost	48.1 (+0.0 points)
Provider reports · OpenAI	63 → 64	Down 1	434 remaining / 0 lost	48.4 (+0.3 points)
All provider reports	63 → 65	Down 2	434 remaining / 0 lost	47.9 (-0.2 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 evidence	Recomputed score	Change
arc-agi	53.4	+5.3
deepswe	43.9	-4.2
mmmu-pro	52.3	+4.2
gdp-pdf	51.5	+3.4
automationbench	44.7	-3.4
itbench	45.8	-2.4
omniscience	49.9	+1.8
hle	49.7	+1.6
aa-lcr	46.7	-1.4
gdpval	46.7	-1.4
critpt	49.5	+1.4
epoch-game-puzzles	47.1	-1.0
terminal-bench-science	49.1	+1.0
briefcase	47.3	-0.8
simpleqa	48.9	+0.8
ifbench	47.3	-0.8
arc-agi-3	47.5	-0.6
frontiercode	47.6	-0.5
tau-bench	47.6	-0.5
enigma-eval	48.5	+0.4
gpqa	47.7	-0.4
terminal-bench	47.8	-0.4
livebench-instructions	48.5	+0.3
multi-swe-bench	47.8	-0.3
mmlu-pro	48.4	+0.3
livebench-language	48.4	+0.3
swe-atlas-qna	47.8	-0.3
apex-agents	48.4	+0.3
enterprise-ops	48.4	+0.2
livecodebench	47.9	-0.2
livebench-coding	47.9	-0.2
aime	47.9	-0.2
analyst-agent	48.3	+0.2
vibe-code	47.9	-0.2
swe-atlas-test-writing	48.0	-0.2
livebench-reasoning	48.0	-0.1
swe-rebench	48.2	+0.1
frontiermath	48.2	+0.1
vals-code-migration	48.0	-0.1
vals-legal-research	48.0	-0.1
swe-atlas-refactoring	48.0	-0.1
math500	48.1	-0.1
vals-tax-agent	48.1	-0.1
harvey	48.1	-0.0
vals-finance-agent	48.1	-0.0
gmmlu	48.1	-0.0
vals-excel-modeling	48.1	-0.0
livebench-data	48.1	+0.0
livebench-math	48.1	+0.0
swe-bench-pro	48.1	+0.0

What evidence is missing from the fit?

Across all collected settings for this model: 27 observations, 23 contributing, 0 matched without graph weight and 4 excluded. Counts do not establish rank eligibility.

2 · Benchmark family or source protocol has not been reviewed
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

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.

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 · 8,360 results across 248 models · Collected 2026-10-02

Benchmark / source	Max	Non-reasoning
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 2026-10-02.	1,420.88 Elo Reported settings DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash	1,116.68 Elo Reported settings DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning
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 2026-10-02.	84% Reported settings DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash	55.33% Reported settings DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning
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 2026-10-02.	68.89% Reported settings DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash	47.81% Reported settings DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning
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. CritPt is under review by its authors following external feedback · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-02.	14.29% Reported settings DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash	0.29% Reported settings DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning
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 2026-10-02.	12.8% Reported settings DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash	3.8% Reported settings DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning
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 2026-10-02.	1,600 Elo Reported settings DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash	1,327.69 Elo Reported settings DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning
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 2026-10-02.	39.25% Reported settings DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash	10.8% Reported settings DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning
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 2026-10-02.	46.92% Reported settings DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash	Not reported
MLCR-AAArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Share of tasks judged accurate, complete, and concise · Expert + compound (hard) sets · Benchmark developed by Wisedocs · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-02.	22.78% Reported settings DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash	Not 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 2026-10-02.	76.99% Reported settings DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash	65.61% Reported settings DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning
AA-Omniscience IndexArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details AA-Omniscience Index (higher is better) measures knowledge reliability and hallucination. It rewards correct answers, penalizes hallucinations, and has no penalty for refusing to answer. Scores range from -100 to 100, where 0 means as many correct as incorrect answers, and negative scores mean more incorrect than correct. · Evaluation results measured independently by Artificial Analysis Unit: index. Source collected 2026-10-02.	-5.3 Reported settings DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash	-10.18 Reported settings DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning
AA-Omniscience AccuracyArtificial Analysis · 6000 questions; 1 attempt; correct-answer share; GPT-5.6 Luna medium grader Benchmark details AA-Omniscience Accuracy (higher is better) measures the proportion of correctly answered questions out of all questions, regardless of whether the model chooses to answer · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-02.	46.4% Reported settings DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash	28.45% Reported settings DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning
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 2026-10-02.	51.85% Reported settings DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash	35.53% Reported settings DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning
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 2026-10-02.	9.05% Reported settings DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash	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 2026-10-02.	26.77% Reported settings DeepSeek V4.1 Flash (Max); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash	5.56% Reported settings DeepSeek V4.1 Flash (Non-reasoning); Model source: https://artificialanalysis.ai/models/deepseek-v4-1-flash-non-reasoning

DeepSeek V4.1 Flash

Compare with

One configuration, seven capabilities

Agentic

Hard reasoning

Coding

Human pref

Knowledge

Multimodal

Long context

Capability score evidence

Benchmark results