Ling 3.1 Flash vs LingSante-3.1-flash
Full profiles: Ling 3.1 Flash (Ant Group) and LingSante-3.1-flash (Ant Group): every benchmark, price and rank.
Updated · Data from 3 sources
As of 22 Nov 2026, there are not enough shared benchmarks for a verdict.
LingSante-3.1-flash was released on 22 Nov 2026 and has no BenchLeader Index yet (it needs five independent results across two categories). So this page gives no quality verdict: it compares what is published so far for both, context window and the benchmarks both have results on. First results are tracked on LingSante-3.1-flash's page.
Key facts (as of 22 Nov 2026)
- Ling 3.1 Flash costs $0.450 per 1M tokens blended (list price, 22 Nov 2026).
- Of the 1 benchmarks both models have, LingSante-3.1-flash leads 1 (BenchLeader, 22 Nov 2026).
Which to pick for…
The pick's numbers come first. Category scores on the BenchLeader scale (50 is the average model, 15 points one standard deviation). High confidence means a lead of half a standard deviation or more; under one point is called either.
| Use case | Pick | Why | Confidence |
|---|---|---|---|
| High volume on a budget | Neither flagship | A smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs | – |
Key differences
- Not enough shared benchmarks for a verdict.
At a glance
| Spec | Ling 3.1 Flash | LingSante-3.1-flash |
|---|---|---|
| Developer | Ant Group | Ant Group |
| Released | 1 Oct 2026 | 22 Nov 2026 |
| Context window (tokens) | 1,000,000 | 1,048,576 |
| Compared at | Ling 3.1 Flash | LingSante-3.1-flash |
| Open weights | Yes | No |
| Price in / out / cached input per 1M | $0.300 / $0.900 / $0.060 | – |
Category by category
Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.
| Category | Ling 3.1 Flash | LingSante-3.1-flash | Leader |
|---|---|---|---|
| Reasoning | 47.3 | – | not comparable |
| Coding | 45.8 | – | not comparable |
| Knowledge | 45.1 | – | not comparable |
| Long context | 66.2 | – | not comparable |
On one chart
Each domain on the index's own scale, where 50 is an average model. Click a domain to see the benchmarks behind it. Pick up to 5 models from the list.
No result on any of these domains: LingSante-3.1-flash.
Not drawn, no result for these models: Agents & tools, Maths, Instruction following, Multimodal, Human preference.
Where sources disagree
Across the 1 benchmarks both have, Ling 3.1 Flash leads 0, LingSante-3.1-flash leads 1.
Each publisher runs its own tasks, harnesses and effort settings, so their overall verdicts can differ; BenchLeader reconciles them on one scale. † measured at another effort setting of the same model (the publisher did not test the one compared here).
Price, speed and latency
List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page. Cost, tokens and time per benchmark task are measured: Artificial Analysis’ averages over its whole Intelligence Index run.
| Measure | Ling 3.1 Flash | LingSante-3.1-flash |
|---|---|---|
| Blended price per 1M tokens | $0.450 | – |
| Input / output per 1M | $0.300 / $0.900 | – / – |
| Cached input per 1M | $0.060 | – |
| Price per task: chat reply | $0.0004 | – |
| Price per task: summarize a 30-page report | $0.0041 | – |
| Price per task: code edit | $0.0032 | – |
| Price per task: agentic coding session | $0.022 | – |
| Price per task: structured extraction | $0.0008 | – |
| Agentic coding session, input mostly cached | $0.011 | – |
| Cost per benchmark task (AA Index) | $3.19 | – |
| Output tokens per task (AA Index) | 225k | – |
| Of which reasoning | 151k | – |
| Output speed | 69 tok/s | – |
| Time to first answer | 45.6 s | – |
| Time to a complete answer, last measured 13 Nov 2026 | 115.5 s | – |
Each model is compared at its best-ranked reasoning-effort setting. Ling 3.1 Flash scores 50.7 and takes 46 s to its first answer token. LingSante-3.1-flash. Higher effort means more thinking before the answer: better scores, longer waits and more tokens billed. How reasoning effort changes scores and cost.
Every benchmark
25 benchmarks, grouped by category; each name links to the board and its leaders. † measured at another effort setting of the same model, shown for reference and not counted as a win.
| Metric | Publisher | Ling 3.1 Flash | LingSante-3.1-flash |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 50.7 | – |
| Coding score | BenchLeader | 45.8 | – |
| Knowledge score | BenchLeader | 45.1 | – |
| Long context score | BenchLeader | 66.2 | – |
| Reasoning score | BenchLeader | 47.3 | – |
| Blended price $/M | BenchLeader | $0.450 | – |
| Output speed | BenchLeader | 69 tok/s | – |
| Time to first answer | BenchLeader | 45.6 s | – |
| Context window | BenchLeader | 1,000,000 tokens | 1,048,576 tokens |
| Reasoning | |||
| Humanity's Last Exam (AA) | Artificial Analysis | 43.5% | – |
| HLE-Diamond (AA) | Artificial Analysis | 17.2% | – |
| CritPt | Artificial Analysis | 26.0% | – |
| CritPt v2 hard subset (AA) | Artificial Analysis | 38.5% | – |
| Coding | |||
| SciCode (AA) | Artificial Analysis | 56.8% | – |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | 33.3% | – |
| GameBench 2 by reasoning setting | DuelLab GameBench 2 | 43.0† | – |
| Design Arena Code | Design Arena | 1296 | – |
| FrontierCode 2.0 (AA) | Artificial Analysis | 17.0% | – |
| Agents & tools | |||
| GDPval-AA v2.1, normalised score | Artificial Analysis | 57.3% | – |
| GDPval-AA v2.1 | Artificial Analysis | 1655 | – |
| AutomationBench | Artificial Analysis | 67.5% | – |
| GDP.pdf (AA) | Artificial Analysis | 12.6% | – |
| AA-Briefcase v1.1 | Artificial Analysis | 1517 | – |
| Design Arena Agentic | Design Arena | 1194 | – |
| Terminal-Bench-Science 0.1 (AA) | Artificial Analysis | 4.3% | – |
| Knowledge | |||
| AA-Omniscience v1.1 | Artificial Analysis | -0.1 | – |
| AA-Omniscience v1.1: accuracy | Artificial Analysis | 26.5% | – |
| AA-Omniscience v1.1: non-hallucination | Artificial Analysis | 63.7% | – |
| Long context | |||
| AA-LCR v1.1 | Artificial Analysis | 83.0% | – |
| MLCR | Artificial Analysis | 15.0% | 23.3% |
| Composite | |||
| AA Intelligence Index v5.0.1 | Artificial Analysis | 29.6 | – |
| AA Intelligence Index v4.3.2 | Artificial Analysis | 43.9 | – |
| AA Progress Index v1 | Artificial Analysis | 321.6 | – |
| Metadata | |||
| AA Openness Index | Artificial Analysis | 38.9 | – |
Data as of 2026-11-22. Best configuration of each model; every score links to its source on the model pages.
DuelLab GameBench 2, release gb2-20261119-1d7fd529faf6 (https://benchmarks.duellab.org/releases/gb2-20261119-1d7fd529faf6/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.
What changed
- LingSante-3.1-flash: first results.
- Ling 3.1 Flash: BenchLeader Index 50.9 → 50.3.
- Ling 3.1 Flash: 1 score revision.
- Ling 3.1 Flash: 1 score revision.
- Ling 3.1 Flash: BenchLeader Index 50.3 → 50.9.
- Ling 3.1 Flash: 6 new results (AA Intelligence Index v5.0.1, AA Progress Index v1, HLE-Diamond (AA)…); 3 score revisions; Ling 3.1 Flash cost per task moved from $1.01/M to $3.19/M (215% up).
- Ling 3.1 Flash: 1 new result (AA-Briefcase v1.1).
- Ling 3.1 Flash: 1 score revision.
- Ling 3.1 Flash: 1 new result (AA-Briefcase v1.1); 8 score revisions.
Related comparisons
Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs
Ling 3.1 Flash vs LingSante-3.1-flash: questions
- Which is better for long documents?
- Both take about 1M tokens: Ling 3.1 Flash 1,000,000, LingSante-3.1-flash 1,048,576.
- Do independent benchmarks agree on Ling 3.1 Flash vs LingSante-3.1-flash?
- Across the 1 benchmarks both have, Ling 3.1 Flash leads 0, LingSante-3.1-flash leads 1.
- When was LingSante-3.1-flash released?
- LingSante-3.1-flash was released on 22 Nov 2026.