Across 48 comparable benchmarks, Claude Fable 5.1 has better reported results on 45 and worse on 2, with 1 tied. 5 further shared results are listed below, each with the reason it is not counted.
Moderate evidence48 of 53 shared results countedHow we compare
DeepSeek-V4.1-Flash is no longer available through its official API. API prices for this version are not shown.
Price basis: Listed rates vs standard rates. The price section below shows the applicable periods.
| Fact | DeepSeek-V4.1-Flash | Claude Fable 5.1 |
|---|---|---|
| Context window | 1M tokens | 1M tokens |
| Input / 1M | โ | $10.00 |
| Output / 1M | โ | $50.00 |
| Blended (4:1) | โ | $10.98 |
| Status | Retired | Available |
| Released | September 10, 2026 | September 1, 2026 |
Blended is 4:1 input to output with 90% cache hits at listed rates โ set your own mix.
Mean of 29 shared benchmarks grouped by capability on a 0โ100 scale. How averages are built
Claude Fable 5.1 ahead on 10 of 12 counted
Claude Fable 5.1 ahead on 3 of 3 counted
Claude Fable 5.1 ahead on 4 of 4 counted
Claude Fable 5.1 ahead on 7 of 7 counted
Claude Fable 5.1 ahead on 1 of 1 counted
No fixed-scale average โ see the rows in Benchmark results
Claude Fable 5.1 ahead on 1 of 1 counted
Claude Fable 5.1 ahead on 5 of 5 counted
Claude Fable 5.1 ahead on 7 of 8 counted
Claude Fable 5.1 ahead on 1 of 1 counted
No shared comparable results for Research & search.
Published scores, with matching modes where available. Gaps show DeepSeek-V4.1-Flash relative to Claude Fable 5.1. Reading the results
| Benchmark | DeepSeek-V4.1-Flash | Claude Fable 5.1 | Gap |
|---|---|---|---|
| AA Cyber IndexRepository Engineering ยท Composite index (0โ100) | 40.8Max ยท tools | 25Max ยท tools | 1.63ร#11 vs #18 of 22 |
| APEX-SWERepository Engineering ยท Pass@1 | 51.9Max ยท tools | 63.6Max ยท tools | โ11.7 pts#12 vs #4 of 61 |
| Code MigrationMaintenance & Optimization ยท Hidden-test pass rate (%) | 45.6Thinking High ยท tools | 54.6โ Max ยท tools | โ9.0 pts#18 vs #12 of 79 |
| DeepSWERepository Engineering ยท Pass@1 (DeepSWE v1.1) | 74.2Max ยท tools | 67.4Max ยท tools | +6.8 pts#11 vs #34 of 73 |
| FrontierCode 1.1Code Generation & Editing ยท Weighted rubric score (%)Not counted ยท Different or unrecorded test environments โ excluded from the tally | 43.9Thinking High ยท toolsยท frontiercode-1-1-main | 63.6โ Thinking Medium ยท toolsยท frontiercode-1-1-extended | โ#18 vs #3 of 51 |
| FrontierCode 2.0Repository Engineering ยท Weighted rubric score | 19.3Max ยท tools + web | 32.4โ Extra-High ยท tools + internet | 0.59ร#22 vs #11 of 36 |
| FrontierCS 2Repository Engineering ยท FCI | 28.1Unspecified Reasoning ยท toolsยท fci | 38.4Unspecified Reasoning ยท toolsยท fci | 0.73ร#13 vs #9 of 21 |
| IOI (Vals v2)Algorithmic Coding ยท Subtask points (%) | 40.3Thinking High ยท tools | 90.8โ Max ยท tools | โ50.5 pts#35 vs #6 of 45 |
| ProgramBenchCode Generation & Editing ยท Almost resolved (>=95%) | 20.3Max ยท toolsยท programbench-almost-resolved | 53.5Max ยท toolsยท programbench-almost-resolved | โ33.2 pts#17 vs #3 of 23 |
| Terminal-Bench 2.1Agentic Development ยท Accuracy | 90.6Max ยท tools | 91.4Max ยท tools | โ0.8 pts#5 vs #3 of 68 |
| Terminal-Bench 4.0Agentic Development ยท Resolution rate (%)Not counted ยท Different or unrecorded test environments โ excluded from the tally | 35.4Max ยท toolsยท vals-effort-study | 58.1โ Unspecified Reasoning ยท toolsยท vals-mini-swe | โ#33 vs #13 of 90 |
| Vibe Code Bench 1-100Agentic Development ยท Accuracy (%) | 16.4Max ยท tools | 28.0Max ยท tools | โ11.6 pts#16 vs #4 of 25 |
| Vibe Code Bench v1.1Code Generation & Editing ยท Pass rate (%) | 84.7Thinking High ยท tools | 90.3โ Max ยท tools | โ5.5 pts#23 vs #8 of 121 |
| WeirdML v3ML Engineering ยท Official score (0-1; higher is better) | 0.059Thinking High ยท tools | 0.26โ Extra-High ยท tools | 0.23ร#22 vs #5 of 26 |
| ARC-AGI-1Abstract Generalization ยท Score (%) | 94.5Max ยท no tools | 97.5Max ยท no tools | โ3.0 pts#21 vs #10 of 87 |
| ARC-AGI-2Abstract Generalization ยท Score (% solved); cost per task (USD)Not counted ยท Speed and cost telemetry is reported, not scored | 72.9Max ยท no tools | 90.0โ Extra-High ยท no tools | 0.81ร#26 vs #10 of 80 |
| GoBenchGeneral Reasoning ยท Elo rating | 1707Max ยท no tools | 2269Max ยท no tools | 0.75ร#11 vs #5 of 18 |
| SimpleBenchCommonsense ยท Score (AVG@5) | 66.7Historical | 86.6โ Thinking ยท no tools | 0.77ร#33 vs #7 of 106 |
| FrontierMath Tier 4 v2Mathematics ยท Accuracy (verification_code) | 26.8Max ยท tools | 87.8Max ยท tools | โ61.0 pts#40 vs #9 of 73 |
| FrontierMath v2Mathematics ยท Accuracy (verification_code) | 67.4Max ยท tools | 90.2Max ยท tools | โ22.8 pts#35 vs #7 of 85 |
| ProofBench v1.1 (Lean 4)Mathematics ยท Accuracy (%) | 54.0Thinking High ยท tools | 100.0โ Max ยท tools | โ46.0 pts#31 vs #1 of 54 |
| Riemann-benchMathematics ยท Accuracy (%) | 24.0Max ยท tools + web | 65.6Max ยท tools + web | โ41.6 pts#25 vs #6 of 42 |
| CritPtScientific Reasoning ยท Score | 14.3Max ยท no tools | 31.1โ Extra-High ยท no tools | 0.46ร#43 vs #8 of 134 |
| GPQA DiamondScientific Reasoning ยท Accuracy | 90.9Max ยท no tools | 93.7Max ยท no tools | โ2.8 pts#44 vs #24 of 200 |
| MysteryMechanismScientific Reasoning ยท Accuracy (%) | 21.2Thinking High ยท tools | 47.8โ Max ยท tools | โ26.6 pts#26 vs #6 of 32 |
| SciCodeScientific Computing ยท Score | 51.9Max ยท no tools | 63.1Max ยท no tools | 0.82ร#43 vs #3 of 98 |
| Terminal-Bench-Science 0.1Scientific Computing ยท Task success rate (%) | 15.7Max ยท tools | 40.5โ Extra-High ยท tools | โ24.8 pts#21 vs #14 of 32 |
| MedCodeClinical Workflows ยท Accuracy (%) | 41.2Thinking High ยท tools | 53.5โ Max ยท tools | โ12.3 pts#54 vs #8 of 97 |
| MedScribeClinical Workflows ยท Rubric pass rate (%) | 85.5Thinking High ยท tools | 91.3โ Max ยท tools | โ5.8 pts#26 vs #2 of 99 |
| APEX-AgentsTool Orchestration ยท Score | 39.5Reported best | 68.6Reported best | 0.58ร#47 vs #6 of 61 |
| AA-LCRLong Reasoning ยท Accuracy | 84.0Max ยท no tools | 85.3Max ยท no tools | โ1.3 pts#8 vs #5 of 22 |
| BabyVisionVisual Understanding ยท Accuracy | 89.6Max ยท tools | 95.1Max ยท tools | โ5.5 pts#10 vs #4 of 17 |
| ChartographyDocuments & Charts ยท Score | 78.9Max ยท tools | 86.2Max ยท tools | 0.92ร#12 vs #6 of 29 |
| GDP.pdfDocuments & Charts ยท Task pass rate (%) | 12.8Max ยท no tools | 28.0โ Thinking Low ยท no tools | โ15.2 pts#50 vs #12 of 96 |
| MMMU-ProVisual Understanding ยท Accuracy | 77.0Max ยท no tools | 88.4โ Max ยท tools | โ11.4 pts#52 vs #4 of 136 |
| ZeroBench MainVisual Understanding ยท Pass@1 / Pass@5 / Pass^5 | 49.0Max ยท tools | 59.0Max ยท tools | โ10.0 pts#10 vs #3 of 21 |
| Creative WritingWriting | 1540Normal ยท no tools | 2155Normal ยท no tools | 0.71ร#54 vs #3 of 121 |
| AutomationBenchOffice & Business ยท Pass RateNot counted ยท Different or unrecorded test environments โ excluded from the tally | 54.8Max ยท toolsยท legacy-default | 31.4โ Max ยท task toolsยท zapier-private-heldout | โ#5 vs #33 of 64 |
| EMB (Excel Modeling)Finance ยท Accuracy (%) | 57.2Thinking High ยท tools | 76.7โ Max ยท tools | โ19.5 pts#42 vs #2 of 78 |
| Finance Agent v2Finance ยท Score | 53.5Thinking High ยท tools | 58.9โ Max ยท tools | 0.91ร#33 vs #8 of 75 |
| Harvey's Legal Agent BenchmarkLegal ยท Accuracy (%) | 6.7Thinking High ยท tools | 6.7โ Max ยท tools | =#21 vs #21 of 52 |
| Legal Research BenchLegal ยท Score | 41.4Thinking High ยท tools | 55.3โ Max ยท tools | 0.75ร#32 vs #3 of 91 |
| Public Benefits Bench v1.1Service Workflows ยท Criterion pass rate (%) | 64.3Thinking High ยท tools | 74.9โ Max ยท tools | โ10.6 pts#23 vs #2 of 50 |
| SAGEService Workflows ยท Accuracy (%) | 47.9Thinking High ยท tools | 48.5โ Max ยท tools | โ0.6 pts#29 vs #27 of 82 |
| Tax Agent BenchFinance ยท Accuracy (%, rubric ร citation quality) | 62.5Thinking High ยท tools | 77.6โ Max ยท tools | โ15.2 pts#15 vs #1 of 74 |
| CWE-bench v1Defensive Remediation ยท Programmatic pass@1 (%) | 55.0Thinking High ยท tools | 58.0Thinking High ยท tools | โ3.0 pts#8 vs #6 of 12 |
| HLEKnowledge Exams ยท Accuracy | 63.9Max ยท tools | 65.6โ Max ยท tools + web | โ1.7 pts#7 vs #4 of 151 |
| Agent ArenaPreference Arenas ยท Net improvement index (percentage points vs. average model) | 2.3Max ยท tools + web | 14.6Max ยท tools + web | 0.15ร#20 vs #6 of 35 |
| ECICapability Indices ยท ECI score (capability index, higher is better) | 155Reported best | 165Reported best | 0.94ร#30 vs #4 of 168 |
| LiveBenchCross-capability Suites ยท Accuracy | 81.1Max ยท no tools | 83.4Max ยท no tools | โ2.3 pts#8 vs #1 of 106 |
| LiveBench 2.0Cross-capability Suites ยท Overall score (equal-weight mean of 7 suite scores, 0-100) | 78.0Max ยท task tools | 88.0Max ยท task tools | 0.89ร#12 vs #1 of 51 |
| Vals IndexCapability IndicesNot counted ยท This metric has no better direction | 57.9Thinking High ยท tools | 68.8โ Max ยท tools | โ |
| Vals Index v2.1Capability Indices ยท GDP-weighted accuracy (%) | 51.3Thinking High ยท tools | 65.8โ Max ยท task tools | โ14.5 pts#23 vs #4 of 38 |
Repository Engineering ยท Composite index (0โ100)
Repository Engineering ยท Pass@1
Maintenance & Optimization ยท Hidden-test pass rate (%)
Repository Engineering ยท Pass@1 (DeepSWE v1.1)
Code Generation & Editing ยท Weighted rubric score (%)
Not counted ยท Different or unrecorded test environments โ excluded from the tally
Repository Engineering ยท Weighted rubric score
Repository Engineering ยท FCI
Algorithmic Coding ยท Subtask points (%)
Code Generation & Editing ยท Almost resolved (>=95%)
Agentic Development ยท Accuracy
Agentic Development ยท Resolution rate (%)
Not counted ยท Different or unrecorded test environments โ excluded from the tally
Agentic Development ยท Accuracy (%)
Code Generation & Editing ยท Pass rate (%)
ML Engineering ยท Official score (0-1; higher is better)
Abstract Generalization ยท Score (%)
Abstract Generalization ยท Score (% solved); cost per task (USD)
Not counted ยท Speed and cost telemetry is reported, not scored
General Reasoning ยท Elo rating
Commonsense ยท Score (AVG@5)
Mathematics ยท Accuracy (verification_code)
Mathematics ยท Accuracy (verification_code)
Mathematics ยท Accuracy (%)
Mathematics ยท Accuracy (%)
Scientific Reasoning ยท Score
Scientific Reasoning ยท Accuracy
Scientific Reasoning ยท Accuracy (%)
Scientific Computing ยท Score
Scientific Computing ยท Task success rate (%)
Clinical Workflows ยท Accuracy (%)
Clinical Workflows ยท Rubric pass rate (%)
Tool Orchestration ยท Score
Long Reasoning ยท Accuracy
Visual Understanding ยท Accuracy
Documents & Charts ยท Score
Documents & Charts ยท Task pass rate (%)
Visual Understanding ยท Accuracy
Visual Understanding ยท Pass@1 / Pass@5 / Pass^5
Writing
Office & Business ยท Pass Rate
Not counted ยท Different or unrecorded test environments โ excluded from the tally
Finance ยท Accuracy (%)
Finance ยท Score
Legal ยท Accuracy (%)
Legal ยท Score
Service Workflows ยท Criterion pass rate (%)
Service Workflows ยท Accuracy (%)
Finance ยท Accuracy (%, rubric ร citation quality)
Defensive Remediation ยท Programmatic pass@1 (%)
Knowledge Exams ยท Accuracy
Preference Arenas ยท Net improvement index (percentage points vs. average model)
Capability Indices ยท ECI score (capability index, higher is better)
Cross-capability Suites ยท Accuracy
Cross-capability Suites ยท Overall score (equal-weight mean of 7 suite scores, 0-100)
Capability Indices
Not counted ยท This metric has no better direction
Capability Indices ยท GDP-weighted accuracy (%)
A dash means no result is available.
Time-of-day providers use peak-hour rates; fixed-price providers use standard rates. Off-peak discounts remain in the published details.
Drag the handle above the bar for cache hits, the one below it for the input : output mix ยท 5% steps
Blended price = effective input rate at 90% cache hits ร 80% + output rate ร 20%
A blended cost needs a published input and output price on both sides, and one of these two does not have them yet.
DeepSeek-V4.1-Flash: Listed text rates; pricing tier not confirmed.
Claude Fable 5.1: Standard pricing. Anthropic official pricing (platform.claude.com/docs/en/about-claude/pricing) verified 2026-10-21; full 1M context at standard price
| Price / 1M tokens | DeepSeek-V4.1-Flash | Claude Fable 5.1 | Ratio |
|---|---|---|---|
| Input | โ | $10.00 | โ |
| Output | โ | $50.00 | โ |
| Cached input | โ | $0.25 | โ |
| Cache write | โ | โ | โ |
| Blended | โ | $10.98 | โ |
Estimates use the selected token mix. Cache writes, tool fees and long-context surcharges are excluded.
Reported cost per task, paired with the score from the same run. Settings and sources may differ; costs do not affect the verdict.
DeepSeek-V4.1-Flash relative to Claude Fable 5.1
Summary: 10 of 13 shared benchmarks have matching recorded settings and sources. The average is a geometric mean of 10 positive-cost ratios, one equal weight per benchmark. This benchmark set can include related tests and indexes; it is not a workload cost estimate.
DeepSeek-V4.1-Flash 247ร cheaper
DeepSeek-V4.1-Flash 217ร cheaper
DeepSeek-V4.1-Flash 53.5ร cheaper
DeepSeek-V4.1-Flash 49ร cheaper
DeepSeek-V4.1-Flash 38.6ร cheaper
DeepSeek-V4.1-Flash 34.7ร cheaper
DeepSeek-V4.1-Flash 34.4ร cheaper
DeepSeek-V4.1-Flash 29.8ร cheaper
DeepSeek-V4.1-Flash 247ร cheaper
DeepSeek-V4.1-Flash 217ร cheaper
DeepSeek-V4.1-Flash 53.5ร cheaper
DeepSeek-V4.1-Flash 49ร cheaper
DeepSeek-V4.1-Flash 38.6ร cheaper
DeepSeek-V4.1-Flash 34.7ร cheaper
DeepSeek-V4.1-Flash 34.4ร cheaper
DeepSeek-V4.1-Flash 29.8ร cheaper
Effort, protocol, aggregation and source match; other run conditions may differ.
DeepSeek-V4.1-Flash
Claude Fable 5.1
Effort, protocol, aggregation and source match; other run conditions may differ.
DeepSeek-V4.1-Flash
Claude Fable 5.1
Effort, protocol, aggregation and source match; other run conditions may differ.
DeepSeek-V4.1-Flash
Claude Fable 5.1
Effort, protocol, aggregation and source match; other run conditions may differ.
DeepSeek-V4.1-Flash
Claude Fable 5.1
Effort, protocol, aggregation and source match; other run conditions may differ.
DeepSeek-V4.1-Flash
Claude Fable 5.1
Effort, protocol, aggregation and source match; other run conditions may differ.
DeepSeek-V4.1-Flash
Claude Fable 5.1
Effort, protocol, aggregation and source match; other run conditions may differ.
DeepSeek-V4.1-Flash
Claude Fable 5.1
Effort, protocol, aggregation and source match; other run conditions may differ.
DeepSeek-V4.1-Flash
Claude Fable 5.1
Settings or sources differ or are not recorded; interpret the comparison with care.
DeepSeek-V4.1-Flash
Claude Fable 5.1
Effort, protocol, aggregation and source match; other run conditions may differ.
DeepSeek-V4.1-Flash
Claude Fable 5.1
Settings or sources differ or are not recorded; interpret the comparison with care.
DeepSeek-V4.1-Flash
Claude Fable 5.1
Effort, protocol, aggregation and source match; other run conditions may differ.
DeepSeek-V4.1-Flash
Claude Fable 5.1
Settings or sources differ or are not recorded; interpret the comparison with care.
DeepSeek-V4.1-Flash
Claude Fable 5.1
| Benchmark | DeepSeek-V4.1-Flash | Claude Fable 5.1 |
|---|---|---|
| AA Coding Agent Index v2.0 | Not reported | |
| AA Intelligence Index v4.3 | Not reported | 2 more settings |
| AA Intelligence Index v5.0.1 | Not reported | 2 more settings |
| AA-Briefcase | Not reported | 2 more settings |
| AA-Omniscience | Not reported | 2 more settings |
| Agent Arena Index | Not reported | $7.7298 / task Median ยท Agent Arena Index rating (index points): 1272 Max (With Tools + Internet) |
| AutomationBench | Not reported | $2.45 / task Average ยท Pass Rate: 31.4 Max (task-specific tools) ยท zapier-private-heldout $2.14 / task Average ยท Pass Rate: 21.3 Extra-High (With Tools) ยท zapier-private-heldout |
| BenchCAD 2.1 Core | Not reported | $34.69 / task Average ยท Core mean geometry/circuit score (0โ100): 44.4 Max (With Tools) $12.44 / task Average ยท Core mean geometry/circuit score (0โ100): 25.8 Thinking High (With Tools) $2.953 / task Average ยท Core mean geometry/circuit score (0โ100): 12.1 Thinking Low (With Tools) 2 more settings$5.703 / task Average ยท Core mean geometry/circuit score (0โ100): 15.7 Thinking Medium (With Tools) $27.38 / task Average ยท Core mean geometry/circuit score (0โ100): 40.5 Extra-High (With Tools) |
| Chartography | Not reported | 1 more settings |
| CursorBench 4.0 | Not reported | $17.28 / task Average ยท Task score: 51.8 Max (With Tools) ยท cursorbench-4-0-cursor-agent $9.08 / task Average ยท Task score: 49.2 Thinking High (With Tools) ยท cursorbench-4-0-cursor-agent $5.44 / task Average ยท Task score: 45.1 Thinking Low (With Tools) ยท cursorbench-4-0-cursor-agent 2 more settings$7.05 / task Average ยท Task score: 46.8 Thinking Medium (With Tools) ยท cursorbench-4-0-cursor-agent $13.01 / task Average ยท Task score: 51.6 Extra-High (With Tools) ยท cursorbench-4-0-cursor-agent |
| FrontierCode 1.1 | Not reported | $12.8257 / task Average ยท Weighted rubric score: 50.3 Max (With Tools) ยท frontiercode-1-1-main $5.2741 / task Average ยท Weighted rubric score: 50.3 Thinking High (With Tools) ยท frontiercode-1-1-main $2.3849 / task Average ยท Weighted rubric score: 49.8 Thinking Low (With Tools) ยท frontiercode-1-1-main 2 more settings$3.2845 / task Average ยท Weighted rubric score: 50.9 Thinking Medium (With Tools) ยท frontiercode-1-1-main $9.2738 / task Average ยท Weighted rubric score: 48.7 Extra-High (With Tools) ยท frontiercode-1-1-main |
| FrontierSWE v2 | Not reported | |
| GDPval-AA v2 | Not reported | 2 more settings |
| GDPx | Not reported | |
| HLE | Not reported | $0.1238 / task Average ยท Accuracy: 48.9 Thinking Low (No Tools) ยท Text only ยท text-only 7 more settings$0.2254 / task Average ยท Accuracy: 53.8 Thinking Medium (No Tools) ยท Text only ยท text-only $0.397 / task Average ยท Accuracy: 55.9 Thinking High (No Tools) ยท Text only ยท text-only $1.0254 / task Average ยท Accuracy: 58.7 Extra-High (No Tools) ยท Text only ยท text-only |
| SWE-sweep | Not reported | |
| Terminal-Bench 4.1 | Not reported | 2 more settings |
| Terminal-Bench-Science 0.1 | Not reported | $38.01 / task Average ยท Task success rate (%): 40.0 Unspecified Reasoning (With Tools) |
| WANDR | Not reported | 2 more settings$28.78 / task Average ยท Soft F1 / relaxed score (%): 64.5 Thinking Medium (With Tools) |
| Attribute | DeepSeek-V4.1-Flash | Claude Fable 5.1 |
|---|---|---|
| Availability | Retired | Available |
| Provider | DeepSeek-AI | Anthropic |
| Model type | Multimodal model | Reasoning model |
| First release | September 10, 2026 | September 1, 2026 |
| Context window | 1M tokens | 1M tokens |
| Max output | 384K tokens | 128K tokens |
| Total parameters | 552B | โ |
| Active parameters | 16B | โ |
| Architecture | Mixture of Experts (MoE) | Undisclosed |
| Documented capabilities | Reasoning, Multilingual, Coding, Function calling, Tool use, Structured output, Long context | Reasoning, Multilingual, Coding, Function calling, Tool use, Computer use, Long context |
| Knowledge cutoff | โ | 2026-06 |
| Input modalities | Text, Image | Text, Image |
| Output modalities | Text | Text |
| Weights | Open weights | Closed / API only |
| Licence | MIT License | โ |
| Commercial use | Commercial use allowed | Proprietary |
These are reported-score comparisons, not controlled head-to-head tests.
What people found comparing these models, or using either one. Comments also appear on the models' own pages.
Used DeepSeek-V4.1-Flash or Claude Fable 5.1? Sign in with Google to share what worked, what did not, and which API or setup you tested.
No community comments on either model yet.
Alternatives with the most benchmark overlap for this matchup.