InclusionAILanguage modelActive

Ling 3.1 Flash

Ling 3.1 Flash has no overall index value yet because there are too few independent benchmark results.

Intelligencenot ranked yet

n/aOverall indexhigher is better

SpeedRank 15 of 101

174Tokens per secondhigher is better

Priceno API price

n/aper 1M tokenslower is better

ContextRank 31 of 102

1Mtokens at once≈ 1,500 A4 pages

PopularityRank 236 of 875

42out of 100+1 vs. last month

Summary

At 174 tokens per second it writes faster than a typical model (87).

Ling 3.1 Flash activates 25 billion of its 560 billion parameters per token. InclusionAI’s hybrid reasoning model handles 262K context tokens on Vercel AI Gateway.

The model card lists MIT for the weights and a 1M-token context for the checkpoint. Vercel AI Gateway offers the model through a separate API route with 262K context tokens.

See for yourself: one prompt, one attempt

SVG short test

One SVG task with the same fixed settings for every model, without hidden thinking.

Ling 3.1 Flash

Earlier tests

API access was unavailable for this run. The model was not evaluated.

Pelican on a bicycle

Prompt

Generate an SVG of a pelican riding a bicycle

Oct 5, 2026

Not yet visually reviewed. A successful recording does not confirm correct physics or full compliance with the prompt.

Test details

Run date: Oct 5, 2026
API provider: openrouter
Requested API model: inclusionai/ling-3.1-flash
Reasoning setting: high
Token limit including reasoning: 8,192 tokens
Pinned endpoint: Novita | inclusionai/ling-3.1-flash-20261002
Temperature: Provider default
Test protocol: community-visual-tests-v1
Technical checks: Not run

Task origin: Simon Willison

All benchmark results

Every published result with its source, test setup and comparison models.

Show all 75 published benchmark results

Measurements without matching peer values

These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.

CritPt v2: Overall (Show measurement, test conditions, and source)

Source value: 18
Score: 18
Metric: pass@1 accuracy
Unit: %
Benchmark version: v2
Category: reasoning
Task: overall
Task label: Overall
Direction: Higher is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: disputed
Retrieved at: 2026-11-02
Methodology: Exact AA configuration: Ling 3.1 Flash (release ling-3-1-flash, row ling-3-1-flash, UUID a6043b9b-628a-4a76-a021-2834e0627d05, reasoning=true, effort=not specified, suffix=not specified). AA source value for average reasoning tokens before the answer: 2000. Its methodology averages 60 varied prompts and assumes 2,000 when no average is available. This is not a token cap. The full CritPt v2 run covers 70 questions with five repeats. The official grading server reports pass@1.. Source model ID: artificial-analysis/a6043b9b-628a-4a76-a021-2834e0627d05/ling-3-1-flash/ling-3-1-flash. Source model name: Ling 3.1 Flash. Provider reported by source: InclusionAI. Model snapshot: Ling 3.1 Flash. Evaluator: Artificial Analysis. Dataset type: artificial-analysis-evaluation. Tool policy: Python functions, symbolic expressions, and numerical answers; official grading server. Attempts: 70 questions, five repeats per question, pass@1. Reasoning: reasoning enabled. Score display: 18.0%. Source ID: artificial-analysis. Evidence capture ID: aa-model-page-2026-11-02-67d46be20a5b. Source payload hash: 67d46be20a5b412db6cfb60ad190441fcaa5513500b8fd47d033168d0e97b6b2. Parser version: artificial-analysis-rsc-models-v3. Imported at: 2026-11-02. Source cohort: artificial-analysis-critpt-v2:v2:standalone:reasoning:effort=unspecified:suffix=unspecified
Comparison methodology: Exact AA configuration: Ling 3.1 Flash (release ling-3-1-flash, row ling-3-1-flash, UUID a6043b9b-628a-4a76-a021-2834e0627d05, reasoning=true, effort=not specified, suffix=not specified). AA source value for average reasoning tokens before the answer: 2000. Its methodology averages 60 varied prompts and assumes 2,000 when no average is available. This is not a token cap. The full CritPt v2 run covers 70 questions with five repeats. The official grading server reports pass@1.. Tool policy: Python functions, symbolic expressions, and numerical answers; official grading server. Attempts: 70 questions, five repeats per question, pass@1. Reasoning: reasoning enabled
Context: This AA source field contains the original CritPt score. The earlier CritPt v2 attribution was incorrect; the value and former label remain as disputed historical evidence.. The source dataset marks this result as disputed. It remains visible as documented evidence, but is not reliable comparative proof.

Source Artificial Analysis

GSM8K completion stop rate (Show measurement, test conditions, and source)

Source value: 99.77
Score: 99.77
Metric: completion stop rate, H200 NEXTN Low-Latency
Unit: %
Direction: No direction published
Source type: independent-evaluator
Evaluator: SGLang Documentation, Ling-3.1-flash cookbook
Status: active
Retrieved at: 2026-10-19
Methodology: Independent evaluation. SGLang GSM8K completion stop-rate figure for the H200 NEXTN Low-Latency recipe. The same section instructs evaluating all 1,319 questions, but the published stop-rate line does not state the completed-run denominator, repeats, or CI.
Context: This is a generation stop/completion-rate check, not GSM8K answer accuracy. The source does not report the completed-run denominator, repeats, or CI for the displayed stop-rate figure.

Source SGLang Documentation, Ling-3.1-flash cookbook

GSM8K completion stop rate (Show measurement, test conditions, and source)

Source value: 99.92
Score: 99.92
Metric: completion stop rate, H200 NEXTN High-Throughput
Unit: %
Direction: No direction published
Source type: independent-evaluator
Evaluator: SGLang Documentation, Ling-3.1-flash cookbook
Status: active
Retrieved at: 2026-10-19
Methodology: Independent evaluation. SGLang GSM8K completion stop-rate figure for the H200 NEXTN High-Throughput recipe. The same section instructs evaluating all 1,319 questions, but the published stop-rate line does not state the completed-run denominator, repeats, or CI.
Context: This is a generation stop/completion-rate check, not GSM8K answer accuracy. The source does not report the completed-run denominator, repeats, or CI for the displayed stop-rate figure.

Source SGLang Documentation, Ling-3.1-flash cookbook

GDPval-AA: Overall (Show measurement, test conditions, and source)

Source value: 1626.38
Score: 1626.38
Metric: GDPval-AA Elo
Unit: Elo
Benchmark version: v2.1
Category: agentic
Task: overall
Task label: Overall
Direction: Higher is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: superseded
Retrieved at: 2026-11-02
Methodology: Exact AA configuration: Ling 3.1 Flash (release ling-3-1-flash, row ling-3-1-flash, UUID a6043b9b-628a-4a76-a021-2834e0627d05, reasoning=true, effort=not specified, suffix=not specified). AA source value for average reasoning tokens before the answer: 2000. Its methodology averages 60 varied prompts and assumes 2,000 when no average is available. This is not a token cap. 220 tasks with one run each. AA scores outcomes through pairwise panel comparisons and anchors the Elo scale at DeepSeek V4.1 Flash (Max) at 1,600.. Source model ID: artificial-analysis/a6043b9b-628a-4a76-a021-2834e0627d05/ling-3-1-flash/ling-3-1-flash. Source model name: Ling 3.1 Flash. Provider reported by source: InclusionAI. Model snapshot: Ling 3.1 Flash. Evaluator: Artificial Analysis. Dataset type: artificial-analysis-evaluation. Harness: Artificial Analysis Stirrup. Agent: Stirrup agent. Tool policy: Agentic task completion with file outputs. Attempts: 220 tasks, one run per task; panel pairwise judgments. Reasoning: reasoning enabled. Source ID: artificial-analysis. Evidence capture ID: aa-model-page-2026-11-02-67d46be20a5b. Source payload hash: 67d46be20a5b412db6cfb60ad190441fcaa5513500b8fd47d033168d0e97b6b2. Parser version: artificial-analysis-rsc-models-v3. Imported at: 2026-11-02. Source cohort: artificial-analysis-gdpval-aa:v2.1:standalone:reasoning:effort=unspecified:suffix=unspecified
Comparison methodology: Exact AA configuration: Ling 3.1 Flash (release ling-3-1-flash, row ling-3-1-flash, UUID a6043b9b-628a-4a76-a021-2834e0627d05, reasoning=true, effort=not specified, suffix=not specified). AA source value for average reasoning tokens before the answer: 2000. Its methodology averages 60 varied prompts and assumes 2,000 when no average is available. This is not a token cap. 220 tasks with one run each. AA scores outcomes through pairwise panel comparisons and anchors the Elo scale at DeepSeek V4.1 Flash (Max) at 1,600.. Harness: Artificial Analysis Stirrup. Agent: Stirrup agent. Tool policy: Agentic task completion with file outputs. Attempts: 220 tasks, one run per task; panel pairwise judgments. Reasoning: reasoning enabled
Context: This result is retained as historical evidence and is not interchangeable with current results.

Source Artificial Analysis

GDPval-AA: Overall (Show measurement, test conditions, and source)

Source value: 1626.56
Score: 1626.56
Metric: GDPval-AA Elo
Unit: Elo
Benchmark version: v2.1
Category: agentic
Task: overall
Task label: Overall
Direction: Higher is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: superseded
Retrieved at: 2026-10-29
Methodology: Exact AA configuration: Ling 3.1 Flash (release ling-3-1-flash, row ling-3-1-flash, UUID a6043b9b-628a-4a76-a021-2834e0627d05, reasoning=true, effort=not specified, suffix=not specified). AA source value for average reasoning tokens before the answer: 2000. Its methodology averages 60 varied prompts and assumes 2,000 when no average is available. This is not a token cap. 220 tasks with one run each. AA scores outcomes through pairwise panel comparisons and anchors the Elo scale at DeepSeek V4.1 Flash (Max) at 1,600.. Source model ID: artificial-analysis/a6043b9b-628a-4a76-a021-2834e0627d05/ling-3-1-flash/ling-3-1-flash. Source model name: Ling 3.1 Flash. Provider reported by source: InclusionAI. Model snapshot: Ling 3.1 Flash. Evaluator: Artificial Analysis. Dataset type: artificial-analysis-evaluation. Harness: Artificial Analysis Stirrup. Agent: Stirrup agent. Tool policy: Agentic task completion with file outputs. Attempts: 220 tasks, one run per task; panel pairwise judgments. Reasoning: reasoning enabled. Source ID: artificial-analysis. Evidence capture ID: aa-model-page-2026-10-29-5a1b1cf41cd1. Source payload hash: 5a1b1cf41cd14d4fa7e1b5c737732505ff7f014395c8a19ff9385c13df09b106. Parser version: artificial-analysis-rsc-models-v3. Imported at: 2026-10-29. Source cohort: artificial-analysis-gdpval-aa:v2.1:standalone:reasoning:effort=unspecified:suffix=unspecified
Comparison methodology: Exact AA configuration: Ling 3.1 Flash (release ling-3-1-flash, row ling-3-1-flash, UUID a6043b9b-628a-4a76-a021-2834e0627d05, reasoning=true, effort=not specified, suffix=not specified). AA source value for average reasoning tokens before the answer: 2000. Its methodology averages 60 varied prompts and assumes 2,000 when no average is available. This is not a token cap. 220 tasks with one run each. AA scores outcomes through pairwise panel comparisons and anchors the Elo scale at DeepSeek V4.1 Flash (Max) at 1,600.. Harness: Artificial Analysis Stirrup. Agent: Stirrup agent. Tool policy: Agentic task completion with file outputs. Attempts: 220 tasks, one run per task; panel pairwise judgments. Reasoning: reasoning enabled
Context: This result is retained as historical evidence and is not interchangeable with current results.

Source Artificial Analysis

GDPval-AA: Overall (Show measurement, test conditions, and source)

Source value: 1626.38
Score: 1626.38
Metric: GDPval-AA Elo
Unit: Elo
Benchmark version: v2.1
Category: agentic
Task: overall
Task label: Overall
Direction: Higher is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: active
Retrieved at: 2026-11-06
Methodology: Exact AA configuration: Ling 3.1 Flash (release ling-3-1-flash, row ling-3-1-flash, UUID a6043b9b-628a-4a76-a021-2834e0627d05, reasoning=true, effort=not specified, suffix=not specified). This source capture does not provide an average reasoning-token count. 220 tasks with one run each. AA scores outcomes through pairwise panel comparisons and anchors the Elo scale at DeepSeek V4.1 Flash (Max) at 1,600.. Source model ID: artificial-analysis/a6043b9b-628a-4a76-a021-2834e0627d05/ling-3-1-flash/ling-3-1-flash. Source model name: Ling 3.1 Flash. Provider reported by source: InclusionAI. Model snapshot: Ling 3.1 Flash. Evaluator: Artificial Analysis. Dataset type: artificial-analysis-evaluation. Harness: Artificial Analysis Stirrup. Agent: Stirrup agent. Tool policy: Agentic task completion with file outputs. Attempts: 220 tasks, one run per task; panel pairwise judgments. Reasoning: reasoning enabled. Source ID: artificial-analysis. Evidence capture ID: aa-model-page-2026-11-06-ec1194e008b3. Source payload hash: ec1194e008b3638ef60dbf5249daf9e6047364e22905e745a4af64c5f3f080fe. Parser version: artificial-analysis-rsc-models-v4. Imported at: 2026-11-06. Source cohort: artificial-analysis-gdpval-aa:v2.1:standalone:reasoning:effort=unspecified:suffix=unspecified
Comparison methodology: Exact AA configuration: Ling 3.1 Flash (release ling-3-1-flash, row ling-3-1-flash, UUID a6043b9b-628a-4a76-a021-2834e0627d05, reasoning=true, effort=not specified, suffix=not specified). This source capture does not provide an average reasoning-token count. 220 tasks with one run each. AA scores outcomes through pairwise panel comparisons and anchors the Elo scale at DeepSeek V4.1 Flash (Max) at 1,600.. Harness: Artificial Analysis Stirrup. Agent: Stirrup agent. Tool policy: Agentic task completion with file outputs. Attempts: 220 tasks, one run per task; panel pairwise judgments. Reasoning: reasoning enabled

Source Artificial Analysis

Artificial Analysis Intelligence Index (Show measurement, test conditions, and source)

Source value: 41.0906200397391
Score: 41.0906200397391
Metric: Artificial Analysis Intelligence Index
Unit: points
Benchmark version: Artificial Analysis Intelligence Index v4.3.2
Category: source-specific
Direction: Higher is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: active
Retrieved at: 2026-10-04
Methodology: Independent evaluation. Artificial Analysis independent evaluation on dedicated hardware; the exact page identifies Ling 3.1 Flash as reasoning-enabled with 2,000 reasoning tokens.
Context: Source-specific observation; it is not a shared comparison cohort.

Source Artificial Analysis

AA-Briefcase (Show measurement, test conditions, and source)

Source value: 1400.35
Score: 1400.35
Metric: AA-Briefcase
Unit: elo
Benchmark version: AA-Briefcase v1.1
Category: source-specific
Direction: Higher is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: active
Retrieved at: 2026-10-04
Methodology: Independent evaluation. 91-task agentic knowledge-work evaluation; the AA page describes a combined Elo over rubric pass rate, analytical-quality Elo and presentation Elo.
Context: Source-specific observation; it is not a shared comparison cohort.

Source Artificial Analysis

GDPval-AA (Show measurement, test conditions, and source)

Source value: 1621.75
Score: 1621.75
Metric: GDPval-AA
Unit: elo
Benchmark version: GDPval-AA v2.1
Category: source-specific
Direction: Higher is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: active
Retrieved at: 2026-10-04
Methodology: Independent evaluation. 220 real-world work tasks, agent loop with shell/web access via Stirrup, blind pairwise comparisons; AA evaluator result for the exact Ling 3.1 Flash slug.
Context: Source-specific observation; it is not a shared comparison cohort.

Source Artificial Analysis

AutomationBench-AA (Show measurement, test conditions, and source)

Source value: 0.6174373352506193
Score: 0.6174373352506193
Metric: AutomationBench-AA
Unit: ratio
Benchmark version: AutomationBench-AA (dataset version 1.0.6 noted on evaluator page)
Category: source-specific
Direction: Higher is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: active
Retrieved at: 2026-10-04
Methodology: Independent evaluation. 657 simulated SaaS workflow tasks across six business domains; exact model result from AA's dedicated-hardware evaluator.
Context: Source-specific observation; it is not a shared comparison cohort.

Source Artificial Analysis

Terminal-Bench (Show measurement, test conditions, and source)

Source value: 0.333333333333333
Score: 0.333333333333333
Metric: Terminal-Bench
Unit: ratio
Benchmark version: Terminal-Bench 4.0
Category: source-specific
Direction: Higher is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: active
Retrieved at: 2026-10-04
Methodology: Independent evaluation. AA page states pass@1 averaged over three repeats per task with the mini-swe-agent harness; exact Ling result.
Context: Source-specific observation; it is not a shared comparison cohort.

Source Artificial Analysis

SciCode (Show measurement, test conditions, and source)

Source value: 0.540509259259259
Score: 0.540509259259259
Metric: SciCode
Unit: ratio
Benchmark version: SciCode current AA implementation
Category: source-specific
Direction: Higher is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: disputed
Retrieved at: 2026-10-04
Methodology: Independent evaluation. 288 test-set subproblems from scientist-curated laboratory problems; independent AA evaluation of the exact checkpoint.
Context: Source-specific observation; it is not a shared comparison cohort. The evaluator currently labels this result Under review.

Source Artificial Analysis

Page 1 of 7

Popularity compared

How often the model is searched for, used and mentioned, compared with every model of its category.

Popularity index

The 100 most popular models of this category, 0 to 100, higher is more popular

20 of 101 models

GPT-573

DeepSeek V4.1 Flash 101768

Kimi K368

DiffusionGemma 26B A4B IT63

Ling 3.1 Flash42

DeepSeek V4.1 Flash 1017

DiffusionGemma 26B A4B IT

Claude Opus 5

Qwen 3.8 Flash

Ling 3.1 Flash

As of November 2026 · Source: gradually.ai popularity index

Over time

Weekly values from October 18, 2026 to November 5, 2026

Ling 3.1 FlashDevstral SmallK2 Horizon 32BNaive-N0.5-FlashGLM-4.6V

All specifications

Architecture

Model class: Hybrid reasoning language model; Source: Model class
Parameters: 560 billion total, 25 billion active; Source: Parameters

Input and output

Context window: 1,000,000 tokens according to the model card; Source: Context window

Watermarking and provenance

AI watermarking refers to invisible signals embedded in the output. Visible marks and provenance metadata are listed separately. Unknown means there is insufficient evidence for that particular field.

AI watermarking: Unknown; no source
Watermarking technique: Not known; no source

Licence and access

Access: Vercel AI Gateway API; Source: Access
License: MIT; Source: License

API

Model ID: inclusionai/ling-3.1-flash; Source: Model ID

Further details

API context window: 262K tokens according to Vercel AI Gateway; Source: API context window
Local inference: Model card documents inference with SGLang and vLLM; Source: Local inference

Prices

No InclusionAI subscription includes this model. Through other providers from ≈ €1.06 a month (Command Code Go, with limited credits). 5 plans in Germany.

Plan	Monthly price	Note
Command Code Go	≈ €1.06	with limited credit
Command Code GOAT	≈ €10.60	with limited credit
Command Code Pro	≈ €21.20	with limited credit
Command Code Max 10×	≈ €106.02	with limited credit
Command Code Max 20×	≈ €212.05	with limited credit

All plans in the AI subscription comparison

Frequently asked questions about Ling 3.1 Flash

How large is the context window of Ling 3.1 Flash?

1 M tokens. That is how much text Ling 3.1 Flash reads at once, about 1,500 A4 pages.

Page count as a rule of thumb from Artificial Analysis

Context windowTokens the model reads at once · higher is better

Ling 3.1 FlashRank 31 of 1021M

Claude Opus 5.51M

Gemini 4 Carbon1M

Claude Fable 5.11M

Claude Sonnet 5.51M

Muse Wonder Full1M

Claude Opus 51M

How fast is Ling 3.1 Flash?

174 tokens per second. That is faster than a typical model at 87.

Measured by Artificial Analysis on Nov 6, 2026. AA 3.0.0, including time to first token, averaged across available workloads and five days.

SpeedTokens per second · higher is better

Gemini 3 Flash Preview191

Gemini 3.5 Flash184

Gemini 3.6 Flash182

DeepSeek-V4.1-Flash176

Ling 3.1 FlashRank 15 of 101174

DeepSeek V4.1 Flash 1017172

Qwen 3.7 Max165

Can I run Ling 3.1 Flash locally?

Yes, the weights are open. With 560B parameters it needs a lot of memory, more a server than a laptop.

Stated by InclusionAI

Open weights: yes
License: MIT
Parameters: 560B
active per token: 25B

Where can I use Ling 3.1 Flash?

Not in a InclusionAI subscription. Including other providers that makes 5 plans in Germany.

Subscription comparison, prices in Germany incl. VAT

At InclusionAI

No InclusionAI subscription includes this model.

Through other providers

5 plans from 1 provider, from ≈ €1.06 a month (Command Code Go), with limited credits.

Sources and data date

Show all 28 sources

Benchmark sources at a glance

Which independent evaluators tested this model. This is coverage, not a quality ranking.

Evaluator	Test series	Composite indices	Recorded active measurements
Artificial Analysis	8	1	10

Without vendor claims and historical or disputed results; those stay labelled in the single values.

Type	Evidence and data date
Speed	InclusionAI, Reasoning enabled; effort unspecified. AA 3.0.0, including time to first token, averaged across available workloads and five days. Retrieved November 6, 2026. Source: Artificial Analysis · Artificial Analysis (retrieved October 4, 2026)Speed measurement methodology
Research status	Research date October 4, 2026. 21 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication.
Unresolved	scicode-and-critpt-evaluation-status Original wording Artificial Analysis labels the SciCode and CritPt observations Under review. They are retained as disputed observations and do not establish confirmed benchmark coverage.
Not published	score scale, denominator, serving revision and confidence interval Note The card does not define chart scales or per-cell sample sizes for these results; no N or CI is inferred.
Not published	GameBench 2 Intensive group Note The source reports: Not supported by this model.
Not published	GameBench 2 Balanced group Note The source reports: No published result.
Additional source	Vercel AI Gateway Ling 3.1 Flash (retrieved October 3, 2026; October 4, 2026) · Vercel AI Gateway Ling 3.1 Flash · Editorial description reviewed October 25, 2026
Additional source	InclusionAI Ling 3.1 Flash model card (retrieved October 17, 2026; October 25, 2026) · InclusionAI Ling 3.1 Flash model card · Editorial description reviewed October 25, 2026
Additional source	Artificial Analysis (retrieved October 4, 2026; November 2, 2026)
Additional source	Artificial Analysis (retrieved October 4, 2026; October 29, 2026; November 2, 2026; November 6, 2026)
Additional source	Artificial Analysis (retrieved October 4, 2026)1 2 3 4 5 6 7 8
Additional source	DataLLM Lab (retrieved October 17, 2026)
Additional source	MathBench AI, Ling-3.1-flash results (retrieved October 17, 2026)
Additional source	SGLang Documentation, Ling-3.1-flash cookbook (retrieved October 19, 2026)
Additional source	DataLLM Lab (retrieved October 17, 2026; October 19, 2026)
Additional source	DuelLab GameBench 2 model results (retrieved October 28, 2026; November 2, 2026; November 5, 2026)
Additional source	Model metadata (retrieved October 4, 2026)
Additional source	Model metadata (retrieved October 17, 2026)1 2
Additional source	OpenRouter: Popularity index (derived), CC BY 4.0
Additional source	Vercel AI Gateway: Popularity index (derived), CC BY 4.0
Additional source	Ollama: cumulative downloads for the popularity index
Additional source	Replicate: cumulative runs for the popularity index
Additional source	Google Trending Now: limited feed selection for Germany and the US
Additional source	Creative Commons Attribution 4.0